Are you UKRM geeks who are still gainfully employed keeping an eye on
what the latest LLMs can do in the planning and coding arena?
If not, heads up, you need to.
On 16/01/2026 08:20, Peter Fisher wrote:
Are you UKRM geeks who are still gainfully employed keeping an eye onIn so many interesting ways it's a mess.
what the latest LLMs can do in the planning and coding arena?
If not, heads up, you need to.
Are you UKRM geeks who are still gainfully employed keeping an eye on
what the latest LLMs can do in the planning and coding arena?
If not, heads up, you need to.
On Fri, 16 Jan 2026 08:20:55 +0000, Peter Fisher <nospam@nosspam.net>
wrote:
Are you UKRM geeks who are still gainfully employed keeping an eye on
what the latest LLMs can do in the planning and coding arena?
If not, heads up, you need to.
I've read lots (on El Reg, etc) about how AI can generate some code
for you, which looks like it works first time, but contains weird bugs
which are almost impossible to trace. And good luck with ever
maintaining it or managing any changes
The internet is filling up with AI slop. Once it starts to fill up
with vibe-coded AI slop systems then <mode=Private Frazer> we're
dooooomed...
On 16/01/2026 17:48, Champ wrote:
On Fri, 16 Jan 2026 08:20:55 +0000, Peter Fisher <nospam@nosspam.net>
wrote:
Are you UKRM geeks who are still gainfully employed keeping an eye on
what the latest LLMs can do in the planning and coding arena?
If not, heads up, you need to.
I've read lots (on El Reg, etc) about how AI can generate some code
for you, which looks like it works first time, but contains weird bugs
which are almost impossible to trace.-a And good luck with ever
maintaining it or managing any changes
The internet is filling up with AI slop.-a Once it starts to fill up
with vibe-coded AI slop systems then <mode=Private Frazer> we're
dooooomed...
Well, I gave Claude a timing system spec. (for the messages it sends) as
a word document and a text file of example raw data from the system
logging. It created VB .net code to read the serial interface, parse the messages and identify which were valid. Took it about a minute. I do a
fair amount of VB code so I could see it was on the right lines, though
it adopted a very different and more modern approach to mine.
You have to give it a plan, though it helped with that by analysing
screen shots of an existing display and asking a few very pertinent
question about what happens at a real speed hill climb. Proposed an architecture which was very much in line with my ideas and took onboard
my insistence on verbose commenting and module resiliance. Also agreed
the need for a local database instance and middleware when we move to updating a database 'somewhere'.
Just one thing it didn't get right first time. Essentially extracting
the split times. It got the count wrong by 1 on first iteration. To be
fair, that was easily trapped because it resulted in a real message
being identified as invalid. It then asked me to show it the code I had
used to do the same task to see my method, 'realised' its error and refactored accordingly.
I shall be exhaustively testing [1] these first modules before getting
it to proceed to catgeorise the messages and allocating times to competitors, but it's looking good so far when I am squirting 'times' at
it from a Qbasic timing box emulator.
So this isn't quite 'vibe coding'.
I hear what you say though. Some very insecure and buggy attempts at
SAAS and all sorts are going to be released in great haste. I think the point is that it will be the donkeywork coders that may lose their jobs,
not Scrum Masters like your goodself.
Like any good workmen we should not blame our tools.
I've read lots (on El Reg, etc) about how AI can generate some code
for you, which looks like it works first time, but contains weird bugs
which are almost impossible to trace. And good luck with ever
maintaining it or managing any changes
The internet is filling up with AI slop. Once it starts to fill up
with vibe-coded AI slop systems then <mode=Private Frazer> we're
dooooomed...
On Jan 16, 2026 at 12:48:59 PM EST, "Champ" <neal@champ.org.uk> wrote:
I've read lots (on El Reg, etc) about how AI can generate some code
for you, which looks like it works first time, but contains weird bugs
which are almost impossible to trace. And good luck with ever
maintaining it or managing any changes
Not my experience. I'm running a project right now where there's lot of code being generated. We ran a blind test, and had the AI generate code for a sprint's worth of user stories, and them compared the code, line-by-line. It was 90% identical. The 10% wa easy to fix.
The internet is filling up with AI slop. Once it starts to fill up
with vibe-coded AI slop systems then <mode=Private Frazer> we're
dooooomed...
Or, an optimistic view: that the AI slop is to generate clicks which generate advertising revenue. When people start not clicking, advertising will fall, AI
content will drop, and we'll get back to quality human-based content...
...such as that great day on UKRM when one of the Father Jacks started a post inviting folks to contribute their favourite euphemism for taking a shit.
On 16/01/2026 14:09, Tim wrote:
On 16/01/2026 08:20, Peter Fisher wrote:
Are you UKRM geeks who are still gainfully employed keeping an eye onIn so many interesting ways it's a mess.
what the latest LLMs can do in the planning and coding arena?
If not, heads up, you need to.
It is, but some recent experimentation with Claude Code Desktop and CLI yielded impressive results. I can see it really being a game changer for
the routinr stuff.
On 16/01/2026 17:21, Peter Fisher wrote:
On 16/01/2026 14:09, Tim wrote:
On 16/01/2026 08:20, Peter Fisher wrote:
Are you UKRM geeks who are still gainfully employed keeping an eyeIn so many interesting ways it's a mess.
on what the latest LLMs can do in the planning and coding arena?
If not, heads up, you need to.
It is, but some recent experimentation with Claude Code Desktop and
CLI yielded impressive results. I can see it really being a game
changer for the routinr stuff.
No junior devs are being hired these days, because they're being
replaced by AI. AI's only useful IMO if you've got senior/experienced
devs to spot and fix the stupid mistkaes it confidently makes.
Where are the next lot of senior devs going to come from?
On Fri, 16 Jan 2026 08:20:55 +0000, Peter Fisher <nospam@nosspam.net>
wrote:
Are you UKRM geeks who are still gainfully employed keeping an eye on
what the latest LLMs can do in the planning and coding arena?
If not, heads up, you need to.
I've read lots (on El Reg, etc) about how AI can generate some code
for you, which looks like it works first time, but contains weird bugs
which are almost impossible to trace. And good luck with ever
maintaining it or managing any changes
The internet is filling up with AI slop. Once it starts to fill up
with vibe-coded AI slop systems then <mode=Private Frazer> we're
dooooomed...
Are you UKRM geeks who are still gainfully employed keeping an eye on
what the latest LLMs can do in the planning and coding arena?
If not, heads up, you need to.
On 16/01/2026 17:48, Champ wrote:
On Fri, 16 Jan 2026 08:20:55 +0000, Peter Fisher <nospam@nosspam.net>
wrote:
Are you UKRM geeks who are still gainfully employed keeping an eye on
what the latest LLMs can do in the planning and coding arena?
If not, heads up, you need to.
I've read lots (on El Reg, etc) about how AI can generate some code
for you, which looks like it works first time, but contains weird bugs
which are almost impossible to trace.-a And good luck with ever
maintaining it or managing any changes
The internet is filling up with AI slop.-a Once it starts to fill up
with vibe-coded AI slop systems then <mode=Private Frazer> we're
dooooomed...
There is a program in place to quickly try to archive as much pre-AI
stuff as we can - https://lowbackgroundsteel.ai/ - the current
generation of LLMs are polluting their source of authoritative
information more quickly than humans can remove it right now.
In article <10kcsd5$1eute$1@dont-email.me>, Peter Fisher says...
Are you UKRM geeks who are still gainfully employed keeping an eye on
what the latest LLMs can do in the planning and coding arena?
If not, heads up, you need to.
This is where it'll lead
https://www.imdb.com/title/tt0387808/
On 16/01/2026 22:24, Neil Ronketti wrote:
There is a program in place to quickly try to archive as much pre-AI
stuff as we can - https://lowbackgroundsteel.ai/ - the current
generation of LLMs are polluting their source of authoritative
information more quickly than humans can remove it right now.
The Arctic Code Vault sounds cool. Svalbard - same place as the Global
Code Vault.
No junior devs are being hired these days, because they're being
replaced by AI. AI's only useful IMO if you've got senior/experienced
devs to spot and fix the stupid mistkaes it confidently makes.
Where are the next lot of senior devs going to come from?
In article <10kcsd5$1eute$1@dont-email.me>, Peter Fisher says...
Are you UKRM geeks who are still gainfully employed keeping an eye on
what the latest LLMs can do in the planning and coding arena?
If not, heads up, you need to.
This is where it'll lead
https://www.imdb.com/title/tt0387808/
On 16/01/2026 21:17 Simon Wilson penned these words:
No junior devs are being hired these days, because they're being
replaced by AI. AI's only useful IMO if you've got senior/experienced
devs to spot and fix the stupid mistkaes it confidently makes.
Where are the next lot of senior devs going to come from?
Once it starts writing its own code, takes control of the robots then it is game
over as the Skynet of the movies becomes real.
On Jan 16, 2026 at 12:48:59 PM EST, "Champ" <neal@champ.org.uk> wrote:
I've read lots (on El Reg, etc) about how AI can generate some code
for you, which looks like it works first time, but contains weird bugs
which are almost impossible to trace. And good luck with ever
maintaining it or managing any changes
Not my experience. I'm running a project right now where there's lot of code being generated. We ran a blind test, and had the AI generate code for a sprint's worth of user stories, and them compared the code, line-by-line. It was 90% identical. The 10% wa easy to fix.
On 16/01/2026 17:48, Champ wrote:
On Fri, 16 Jan 2026 08:20:55 +0000, Peter Fisher <nospam@nosspam.net>
wrote:
Are you UKRM geeks who are still gainfully employed keeping an eye on
what the latest LLMs can do in the planning and coding arena?
If not, heads up, you need to.
I've read lots (on El Reg, etc) about how AI can generate some code
for you, which looks like it works first time, but contains weird bugs
which are almost impossible to trace.-a And good luck with ever
maintaining it or managing any changes
The internet is filling up with AI slop.-a Once it starts to fill up
with vibe-coded AI slop systems then <mode=Private Frazer> we're
dooooomed...
There is a program in place to quickly try to archive as much pre-AI
stuff as we can - https://lowbackgroundsteel.ai/ - the current
generation of LLMs are polluting their source of authoritative
information more quickly than humans can remove it right now.
On 16/01/2026 22:24, Neil Ronketti wrote:
On 16/01/2026 17:48, Champ wrote:Would that be a programme or program, just for clarity and this is just
On Fri, 16 Jan 2026 08:20:55 +0000, Peter Fisher <nospam@nosspam.net>
wrote:
Are you UKRM geeks who are still gainfully employed keeping an eye on
what the latest LLMs can do in the planning and coding arena?
If not, heads up, you need to.
I've read lots (on El Reg, etc) about how AI can generate some code
for you, which looks like it works first time, but contains weird bugs
which are almost impossible to trace.-a And good luck with ever
maintaining it or managing any changes
The internet is filling up with AI slop.-a Once it starts to fill up
with vibe-coded AI slop systems then <mode=Private Frazer> we're
dooooomed...
There is a program in place to quickly try to archive as much pre-AI
stuff as we can - https://lowbackgroundsteel.ai/ - the current
generation of LLMs are polluting their source of authoritative
information more quickly than humans can remove it right now.
the kind of thing the A.I. trips over.
On 16/01/2026 23:08 petrolcan penned these words:
In article <10kcsd5$1eute$1@dont-email.me>, Peter Fisher says...
Are you UKRM geeks who are still gainfully employed keeping an eye on
what the latest LLMs can do in the planning and coding arena?
If not, heads up, you need to.
This is where it'll lead
https://www.imdb.com/title/tt0387808/
I fear we're already there with the "drill baby drill"
It is, but some recent experimentation with Claude Code Desktop and CLI yielded impressive results. I can see it really being a game changer for
the routinr stuff.
On 16/01/2026 17:21, Peter Fisher wrote:
It is, but some recent experimentation with Claude Code Desktop and
CLI yielded impressive results. I can see it really being a game
changer for the routinr stuff.
It's quite good for new code. But if you ask it to modify something
existing it's pretty poor at the moment.
On 16/01/2026 17:21, Peter Fisher wrote:
It is, but some recent experimentation with Claude Code Desktop and
CLI yielded impressive results. I can see it really being a game
changer for the routinr stuff.
It's quite good for new code. But if you ask it to modify something
existing it's pretty poor at the moment.
On 17/01/2026 12:15, Tim wrote:
On 16/01/2026 22:24, Neil Ronketti wrote:
On 16/01/2026 17:48, Champ wrote:Would that be a programme or program, just for clarity and this is
On Fri, 16 Jan 2026 08:20:55 +0000, Peter Fisher <nospam@nosspam.net>
wrote:
Are you UKRM geeks who are still gainfully employed keeping an eye on >>>>> what the latest LLMs can do in the planning and coding arena?
If not, heads up, you need to.
I've read lots (on El Reg, etc) about how AI can generate some code
for you, which looks like it works first time, but contains weird bugs >>>> which are almost impossible to trace.-a And good luck with ever
maintaining it or managing any changes
The internet is filling up with AI slop.-a Once it starts to fill up
with vibe-coded AI slop systems then <mode=Private Frazer> we're
dooooomed...
There is a program in place to quickly try to archive as much pre-AI
stuff as we can - https://lowbackgroundsteel.ai/ - the current
generation of LLMs are polluting their source of authoritative
information more quickly than humans can remove it right now.
just the kind of thing the A.I. trips over.
Two nations divided by a common language.
Are you UKRM geeks who are still gainfully employed keeping an eye on
what the latest LLMs can do in the planning and coding arena?
If not, heads up, you need to.
On 17/01/2026 12:31, Peter Fisher wrote:
On 17/01/2026 12:15, Tim wrote:
On 16/01/2026 22:24, Neil Ronketti wrote:
On 16/01/2026 17:48, Champ wrote:Would that be a programme or program, just for clarity and this is
On Fri, 16 Jan 2026 08:20:55 +0000, Peter Fisher <nospam@nosspam.net> >>>>> wrote:
Are you UKRM geeks who are still gainfully employed keeping an eye on >>>>>> what the latest LLMs can do in the planning and coding arena?
If not, heads up, you need to.
I've read lots (on El Reg, etc) about how AI can generate some code
for you, which looks like it works first time, but contains weird bugs >>>>> which are almost impossible to trace.-a And good luck with ever
maintaining it or managing any changes
The internet is filling up with AI slop.-a Once it starts to fill up >>>>> with vibe-coded AI slop systems then <mode=Private Frazer> we're
dooooomed...
There is a program in place to quickly try to archive as much pre-AI
stuff as we can - https://lowbackgroundsteel.ai/ - the current
generation of LLMs are polluting their source of authoritative
information more quickly than humans can remove it right now.
just the kind of thing the A.I. trips over.
Two nations divided by a common language.
True to an extent, but common usage is that "programme" is a plan/
behaviour and "program" is code.
On 16/01/2026 08:20, Peter Fisher wrote:
Are you UKRM geeks who are still gainfully employed keeping an eye on
what the latest LLMs can do in the planning and coding arena?
If not, heads up, you need to.
Given their recent record in inventing legal case history and football matches, should they really be trusted?
I still wouldn't trust it to code mission-critical software.
On Jan 18, 2026 at 10:50:49 AM EST, "Peter Fisher" <nospam@nosspam.net> wrote:
I still wouldn't trust it to code mission-critical software.
Nor would I, if I were planning to deploy without any testing.
But with well-written acceptance criteria for every user story,
and with
sufficient time for detailed and careful human testing (plus additive automated testing for monte carlo type scenarios), for SIT, UAT, and Hypercare, I don't see why I wouldn't trust AI generated mission-critical code.
On Jan 18, 2026 at 10:50:49 AM EST, "Peter Fisher" <nospam@nosspam.net> wrote:
I still wouldn't trust it to code mission-critical software.
Nor would I, if I were planning to deploy without any testing.
But with well-written acceptance criteria for every user story, and with sufficient time for detailed and careful human testing (plus additive automated testing for monte carlo type scenarios), for SIT, UAT, and Hypercare, I don't see why I wouldn't trust AI generated mission-critical code.
On 18/01/2026 18:06, Ben Blaney wrote:
On Jan 18, 2026 at 10:50:49 AM EST, "Peter Fisher" <nospam@nosspam.net> wrote:
I still wouldn't trust it to code mission-critical software.
Nor would I, if I were planning to deploy without any testing.
But with well-written acceptance criteria for every user story,
AI generated?
and with
sufficient time for detailed and careful human testing (plus additive
automated testing for monte carlo type scenarios), for SIT, UAT, and
Hypercare, I don't see why I wouldn't trust AI generated mission-critical
code.
They did used to say that every program has on average a bug for every N lines of code, though I don't recall what N was.
They did used to say that every program has on average a bug for every N lines of code, though I don't recall what N was.
Are you UKRM geeks who are still gainfully employed keeping an eye on
what the latest LLMs can do in the planning and coding arena?
If not, heads up, you need to.
But with well-written acceptance criteria for every user story, and with sufficient time for detailed and careful human testing (plus additive automated testing for monte carlo type scenarios), for SIT, UAT, and Hypercare, I don't see why I wouldn't trust AI generated mission-critical code.
On 18/01/2026 18:06, Ben Blaney wrote:
But with well-written acceptance criteria for every user story, and with
sufficient time for detailed and careful human testing (plus additive
automated testing for monte carlo type scenarios), for SIT, UAT, and
Hypercare, I don't see why I wouldn't trust AI generated mission-critical
code.
Jeezus, I am so glad I don't do this stuff anymore.
On 16/01/2026 08:20, Peter Fisher wrote:
Are you UKRM geeks who are still gainfully employed keeping an eye on
what the latest LLMs can do in the planning and coding arena?
If not, heads up, you need to.
Some interesting points in this article:
https://webmatrices.com/post/vibe-coding-has-a-12x-cost-problem- maintainers-are-done
tl;dr
AI-assisted coding works great for:
personal projects
prototypes
learning
throwaway scripts
experienced developers who can review what they generate
AI-assisted coding destroys:
collaborative codebases (when contributors can't review their own output) open source projects (when maintainers bear the review cost)
anything someone else has to debug
anything that needs to be secure
On 18/01/2026 18:06, Ben Blaney wrote:
But with well-written acceptance criteria for every user story, and with
sufficient time for detailed and careful human testing (plus additive
automated testing for monte carlo type scenarios), for SIT, UAT, and
Hypercare, I don't see why I wouldn't trust AI generated mission-critical
code.
Jeezus, I am so glad I don't do this stuff anymore.
On 19/01/2026 22:16, Peter Fisher wrote:
On 19/01/2026 14:50, Simon Wilson wrote:This one was developed using AI, apparently.
On 16/01/2026 08:20, Peter Fisher wrote:
Are you UKRM geeks who are still gainfully employed keeping an eye
on what the latest LLMs can do in the planning and coding arena?
If not, heads up, you need to.
Some interesting points in this article:
https://webmatrices.com/post/vibe-coding-has-a-12x-cost-problem-
maintainers-are-done
tl;dr
AI-assisted coding works great for:
personal projects
Right up UKRM's street in many ways.
https://www.ismypubfucked.com/
On 20/01/2026 07:42, Peter Fisher wrote:
This one was developed using AI, apparently.
Right up UKRM's street in many ways.
https://www.ismypubfucked.com/
On 20/01/2026 16:31, Tim wrote:
On 20/01/2026 07:42, Peter Fisher wrote:
This one was developed using AI, apparently.
Right up UKRM's street in many ways.
https://www.ismypubfucked.com/
My local is 'feeling it' under the terms of this - but so are all the
others within a mile - or are worse off
:-(
Chris
On 20/01/2026 16:40, chrisnd @ukrm wrote:
On 20/01/2026 16:31, Tim wrote:
On 20/01/2026 07:42, Peter Fisher wrote:
This one was developed using AI, apparently.
Right up UKRM's street in many ways.
https://www.ismypubfucked.com/
My local is 'feeling it' under the terms of this - but so are all the
others within a mile - or are worse off
:-(
Chris
Same here, except one, which is 'Somehow Fine'. It's been closed for
years. Never got permission for change of use to housing and is now undergoing refurbishment. I suspect they will find it hard to make a go
of it.
Peter Fisher <nospam@nosspam.net> wrote in news:10kos9n$1j5bc$1@dont- email.me:
On 20/01/2026 16:40, chrisnd @ukrm wrote:
On 20/01/2026 16:31, Tim wrote:
On 20/01/2026 07:42, Peter Fisher wrote:
This one was developed using AI, apparently.
Right up UKRM's street in many ways.
https://www.ismypubfucked.com/
My local is 'feeling it' under the terms of this - but so are all the
others within a mile - or are worse off
:-(
Chris
Same here, except one, which is 'Somehow Fine'. It's been closed for
years. Never got permission for change of use to housing and is now
undergoing refurbishment. I suspect they will find it hard to make a go
of it.
a 50-50 mix here of "fine" and "feeling it"
some massive differences in the projected rate increase, from 2 to 35%. Not sure where they got the figures from as Paul in my preferred local pub
states he has yet to find out what his new business rates will be.
The one that will struggle is the one owned by Punch Taverns with a new tenant. Punch will squeeze the tenant until his pips squeak.
As long as Paul does not increase the price too much of a pint of Station Bitter from the Stonehouse Brewery in Oswestry. Currently +U3.80 a pint. https://www.stonehousebrewery.co.uk/trade/
On 19/01/2026 22:16, Peter Fisher wrote:
This one was developed using AI, apparently.
Right up UKRM's street in many ways.
https://www.ismypubfucked.com/
On 20/01/2026 07:42 Peter Fisher penned these words:
On 19/01/2026 22:16, Peter Fisher wrote:
This one was developed using AI, apparently.
Right up UKRM's street in many ways.
https://www.ismypubfucked.com/
The one closest to me seems to be misnamed and where the school is,
not sure on the accuracy of their data.
boots <news@millhouse-communications.co.uk> wrote in news:10kq84n$20dvb$2@bilbo.eternal-september.org:
On 20/01/2026 07:42 Peter Fisher penned these words:
On 19/01/2026 22:16, Peter Fisher wrote:
This one was developed using AI, apparently.
Right up UKRM's street in many ways.
https://www.ismypubfucked.com/
The one closest to me seems to be misnamed and where the school is,
not sure on the accuracy of their data.
one of my local pubs is showing under its old pre 2023 name as a Bistro. Maybe the old trading name is still what the business is registered as with the council and therefore the database they used?
On 21/01/2026 11:35 wessie penned these words:
boots <news@millhouse-communications.co.uk> wrote in
news:10kq84n$20dvb$2@bilbo.eternal-september.org:
On 20/01/2026 07:42 Peter Fisher penned these words:
On 19/01/2026 22:16, Peter Fisher wrote:
This one was developed using AI, apparently.
Right up UKRM's street in many ways.
https://www.ismypubfucked.com/
The one closest to me seems to be misnamed and where the school is,
not sure on the accuracy of their data.
one of my local pubs is showing under its old pre 2023 name as a Bistro.
Maybe the old trading name is still what the business is registered as with >> the council and therefore the database they used?
Could be we've not been here long enough, it is in the wrong place too but close
"...then join the datasets using the UARN (Unique Address Reference Number)."
I used to work with similar UPRNs back when I was developing a pest
control service diary system. Supplied by the authority's digital
mapping officer. Garbage in... I suspect the national version may be as inaccurate as some of Google Maps locations.
Peter Fisher <nospam@nosspam.net> wrote:
"...then join the datasets using the UARN (Unique Address Reference
Number)."
I used to work with similar UPRNs back when I was developing a pest
control service diary system. Supplied by the authority's digital
mapping officer. Garbage in... I suspect the national version may be as
inaccurate as some of Google Maps locations.
Every day's a school day. Google found this article:
https://www.geoplace.co.uk/blog/persistent-and-well-behaved-identifiers
It lists a few similar acronyms (or initialisms, depending on how
you say them):
MPAN, MPRN, UARN, UPRN, USRN and probably a couple more I missed.
A couple of points I thought were well made:
"When these links are used for important decisions and services,
such as mobilising a fire engine or switching energy providers,
errors can be serious and irreversible."
"Links should not be made without due care and it is better not to
make a link than to make an incorrect one."
Once it starts writing its own code, takes control of the robots then it is game
over as the Skynet of the movies becomes real.
On 18/01/2026 18:06, Ben Blaney wrote:
But with well-written acceptance criteria for every user story, and with
sufficient time for detailed and careful human testing (plus additive
automated testing for monte carlo type scenarios), for SIT, UAT, and
Hypercare, I don't see why I wouldn't trust AI generated mission-critical
code.
Jeezus, I am so glad I don't do this stuff anymore.
On Sat, 17 Jan 2026 08:45:01 +0000, boots <news@millhouse-communications.co.uk> wrote:
Once it starts writing its own code, takes control of the robots then it is game
over as the Skynet of the movies becomes real.
That's not going to happen. LLMs have no actual intelligence at all.
They're just predictive text in a posh frock
On Sat, 17 Jan 2026 08:45:01 +0000, boots <news@millhouse-communications.co.uk> wrote:
Once it starts writing its own code, takes control of the robots then it is game
over as the Skynet of the movies becomes real.
That's not going to happen. LLMs have no actual intelligence at all.
They're just predictive text in a posh frock
On Sat, 17 Jan 2026 08:45:01 +0000, boots <news@millhouse-communications.co.uk> wrote:
Once it starts writing its own code, takes control of the robots then it is game
over as the Skynet of the movies becomes real.
That's not going to happen. LLMs have no actual intelligence at all.
They're just predictive text in a posh frock
On 22/01/2026 08:57, Champ wrote:
On Sat, 17 Jan 2026 08:45:01 +0000, boots
<news@millhouse-communications.co.uk> wrote:
Once it starts writing its own code, takes control of the robots then it is game
over as the Skynet of the movies becomes real.
That's not going to happen. LLMs have no actual intelligence at all.
They're just predictive text in a posh frock
Very far from AGI, agreed. I really do think, however, you ought to try interacting with a system planning/coding specific model like Claude
Code. As you are no longer in the game, it is of no consequence to you,
but arguing from a position of ignorance is never a good look.
Peter Fisher <nospam@nosspam.net> wrote:
On 22/01/2026 08:57, Champ wrote:
On Sat, 17 Jan 2026 08:45:01 +0000, boots
<news@millhouse-communications.co.uk> wrote:
Once it starts writing its own code, takes control of the robots then it is game
over as the Skynet of the movies becomes real.
That's not going to happen. LLMs have no actual intelligence at all.
They're just predictive text in a posh frock
Very far from AGI, agreed. I really do think, however, you ought to try
interacting with a system planning/coding specific model like Claude
Code. As you are no longer in the game, it is of no consequence to you,
but arguing from a position of ignorance is never a good look.
What really bothers me about the whole thing is how easy it is for
people involved to uncritically adopt the idea that it is OK and
normal that they should ascribe human emotions and motives to LLMs.
Please note that I am not lumping Mr. Fisher into this category.
A recent El Reg article seems appropriate to bring up at this point.
https://www.theregister.com/2026/01/22/anthropic_claude_constitution/
What really bothers me about the whole thing is how easy it is for
people involved to uncritically adopt the idea that it is OK and
normal that they should ascribe human emotions and motives to LLMs.
As Weizenbaum
later wrote, 'I had not realized ... that extremely short exposures to
a relatively simple computer program could induce powerful delusional thinking in quite normal people.'"
+ https://en.wikipedia.org/wiki/ELIZA_effect
As others have said, I'm glad to be out of the business and I have
the luxury of choosing (for a little while at least?) to not be
forced to use AI. However, as AI becomes more and more woven into everything, I fear we will not have much of a choice in the future.
Yet again, lo! the soul, above all science,--
For it has history gather'd like husks around the globe,
For it the entire star-myriads roll through the sky.
In spiral routes by long detours,
(As a much-tacking ship upon the sea,)
For it the partial to the permanent flowing,
For it the real to the ideal tends.
Excerpt from "Song of the Universal" by Walt Whitman (1819-1892)
On Thu, 22 Jan 2026 23:03:10 -0000 (UTC), Mark Olson <olsonm@tiny.invalid> wrote:
As others have said, I'm glad to be out of the business and I have
the luxury of choosing (for a little while at least?) to not be
forced to use AI. However, as AI becomes more and more woven into
everything, I fear we will not have much of a choice in the future.
These days there's a toad under every rock waiting to hop out and tell
you that you can't avoid the enshittification of AI. Its all over Big
Media. Instead I think you need to cultivate your sources in the counterculture. The trouble is the toads are there, too.
On 24/01/2026 00:26, "Worst Case"@dizum.com wrote:
On Thu, 22 Jan 2026 23:03:10 -0000 (UTC), Mark Olson
<olsonm@tiny.invalid> wrote:
As others have said, I'm glad to be out of the business and I have
the luxury of choosing (for a little while at least?) to not be
forced to use AI.-a However, as AI becomes more and more woven into
everything, I fear we will not have much of a choice in the future.
These days there's a toad under every rock waiting to hop out and tell
you that you can't avoid the enshittification of AI.-a Its all over Big
Media.-a Instead I think you need to cultivate your sources in the
counterculture.-a The trouble is the toads are there, too.
<snipola>
I'm sure I saw Toads and the Counterculture supporting The Fall sometime
in about 1983.
On Thu, 22 Jan 2026 23:03:10 -0000 (UTC), Mark Olson <olsonm@tiny.invalid> wrote:
As others have said, I'm glad to be out of the business and I have
the luxury of choosing (for a little while at least?) to not be
forced to use AI. However, as AI becomes more and more woven into
everything, I fear we will not have much of a choice in the future.
These days there's a toad under every rock waiting to hop out and tell
you that you can't avoid the enshittification of AI. Its all over Big
Media. Instead I think you need to cultivate your sources in the counterculture. The trouble is the toads are there, too.
Yet again, lo! the soul, above all science,
For it has history gather'd like husks around the globe,
For it the entire star-myriads roll through the sky.
In spiral routes by long detours,
(As a much-tacking ship upon the sea,)
For it the partial to the permanent flowing,
For it the real to the ideal tends.
Excerpt from "Song of the Universal" by Walt Whitman (1819-1892)
On 22/01/2026 08:57, Champ wrote:
On Sat, 17 Jan 2026 08:45:01 +0000, boots
<news@millhouse-communications.co.uk> wrote:
Once it starts writing its own code, takes control of the robots then it is game
over as the Skynet of the movies becomes real.
That's not going to happen. LLMs have no actual intelligence at all.
They're just predictive text in a posh frock
Very far from AGI, agreed. I really do think, however, you ought to try >interacting with a system planning/coding specific model like Claude
Code. As you are no longer in the game, it is of no consequence to you,
but arguing from a position of ignorance is never a good look.
That's not going to happen. LLMs have no actual intelligence at all.
They're just predictive text in a posh frock
They do pass the Turing test I believe, but does that mean they are >"intelligent"? Sometimes it can take a while to realise one is talking
to a bot.
On Thu, 22 Jan 2026 10:33:24 +0000, Peter Fisher <nospam@nosspam.net>
wrote:
On 22/01/2026 08:57, Champ wrote:
On Sat, 17 Jan 2026 08:45:01 +0000, boots
<news@millhouse-communications.co.uk> wrote:
Once it starts writing its own code, takes control of the robots then it is game
over as the Skynet of the movies becomes real.
That's not going to happen. LLMs have no actual intelligence at all.
They're just predictive text in a posh frock
Very far from AGI, agreed. I really do think, however, you ought to try
interacting with a system planning/coding specific model like Claude
Code. As you are no longer in the game, it is of no consequence to you,
but arguing from a position of ignorance is never a good look.
I strongly refute that I'm arguing from a position of ignorance.
Claude can obviously produce working code, but it's fooling you if you
think this is any more than stastical processing of all the other code
that's already out there.
There are apparently plenty of people out there with AI boyfriends/girlfriends. These people are genuinely attached to the
LLM they interface with. That's how these things work - you interact
with it, and it's so believable you can't accept that it's just stats.
But it is
But I'm not aware of anyone working in the implementation of LLMs
claiming they are actually thinking
I'm also pretty uncomfortable sitting here wondering what to do after
the Gestapo/Brown Shirts v2.0 have just blatantly and transparently
murdered another person about 20 miles from me.
On 24/01/2026 17:54, Mark Olson wrote:
I'm also pretty uncomfortable sitting here wondering what to do after
the Gestapo/Brown Shirts v2.0 have just blatantly and transparently
murdered another person about 20 miles from me.
I was gonna ask about your state. Seems rather... worrying.
Also it seems some people believe that the right to bear arms was only intended for right wing/MAGA white people.
It must be really scary. I'm sorry. Seems there's very little I can do
other than try to avoid buying anything made in the USA. Or travelling there.
On 24/01/2026 17:54, Mark Olson wrote:
I'm also pretty uncomfortable sitting here wondering what to do after
the Gestapo/Brown Shirts v2.0 have just blatantly and transparently
murdered another person about 20 miles from me.
I was gonna ask about your state. Seems rather... worrying.
Also it seems some people believe that the right to bear arms was only intended for right wing/MAGA white people.
It must be really scary. I'm sorry. Seems there's very little I can do
other than try to avoid buying anything made in the USA. Or travelling there.
Champ <neal@champ.org.uk> wrote:
But I'm not aware of anyone working in the implementation of LLMs
claiming they are actually thinking
I'm less worried about the thought processes of people working in the implementation of LLMs (although there IS plenty to worry about there)
than I am about the vastly larger number of others who are rushing
headlong toward blindly trusting the output of same, and uncritically anthropomorphizing them.
I'm also pretty uncomfortable sitting here wondering what to do after
the Gestapo/Brown Shirts v2.0 have just blatantly and transparently
murdered another person about 20 miles from me.
On 24/01/2026 17:54, Mark Olson wrote:
I'm also pretty uncomfortable sitting here wondering what to do after
the Gestapo/Brown Shirts v2.0 have just blatantly and transparently
murdered another person about 20 miles from me.
Must be a cause for justified anxiety. Given the social media screening,
I don't think I would be let in to the states now, even if I wanted to
go across to call in on 'Zot' to put the world to rights, as we did the
last time he stayed at our gaff.
Peter Fisher <nospam@nosspam.net> wrote:
On 24/01/2026 17:54, Mark Olson wrote:
I'm also pretty uncomfortable sitting here wondering what to do after
the Gestapo/Brown Shirts v2.0 have just blatantly and transparently
murdered another person about 20 miles from me.
Must be a cause for justified anxiety. Given the social media screening,
I don't think I would be let in to the states now, even if I wanted to
go across to call in on 'Zot' to put the world to rights, as we did the
last time he stayed at our gaff.
I wonder how our Thomas is doing lately. I'm not on Farcebook and I
haven't seen him on here much. I have his mobile number from 2013
when a buddy and I dropped in on him for an overnight visit, so I
suppose I really have no excuse not to simply text him.
The missus and I have a European trip (Munich & Lubljuna|Maribor)
booked a couple of months from now and I am starting to get worried
about how we will be received. Maybe we should put a Minnesota patch
on our jackets, akin to the old ruse of sewing a Canadian flag on
one's backpack so as not to be taken for an ugly American. I don't
think I could bear the shame of pretending to be a Canadian for my
own benefit after our administration has done so much to insult and
indeed threaten them.
I suppose by openly posting my sentiments here under my own name,
I am inviting scrutiny at US customs, too.
Simon Wilson <siwilson@nodamnspamn.hotmail.com> wrote:
On 24/01/2026 17:54, Mark Olson wrote:
I'm also pretty uncomfortable sitting here wondering what to do after
the Gestapo/Brown Shirts v2.0 have just blatantly and transparently
murdered another person about 20 miles from me.
I was gonna ask about your state. Seems rather... worrying.
Also it seems some people believe that the right to bear arms was only
intended for right wing/MAGA white people.
No lie there.
It must be really scary. I'm sorry. Seems there's very little I can do
other than try to avoid buying anything made in the USA. Or travelling
there.
Thanks for the reply and concern.
I have a daughter that lives a couple of blocks from there. My daughter-in-law is a US citizen, born in Seoul, of Korean-American
parentage. She's afraid to come down near Minneapolis because she might
be perceived as an immigrant and detained, despite being a citizen.
Events have shown that _no one_ is safe from arbitrary detention,
arrest, disappearance, or even outright execution in public.
On Jan 24, 2026 at 1:19:52 PM EST, "Mark Olson" <olsonm@tiny.invalid> wrote:
Simon Wilson <siwilson@nodamnspamn.hotmail.com> wrote:
On 24/01/2026 17:54, Mark Olson wrote:
I'm also pretty uncomfortable sitting here wondering what to do after
the Gestapo/Brown Shirts v2.0 have just blatantly and transparently
murdered another person about 20 miles from me.
Fucking awful, isn't it. I don't know how this ends.
https://www.theguardian.com/us-news/live/2026/jan/24/minneapolis-shooting-live?CMP=share_btn_url&page=with%3Ablock-697523318f0839147949025b#block-697523318f0839147949025b--
Not looking exactly like a peace and love scenario:
https://www.theguardian.com/us-news/live/2026/jan/24/minneapolis-shooting-live
Are you UKRM geeks who are still gainfully employed keeping an eye on
what the latest LLMs can do in the planning and coding arena?
If not, heads up, you need to.
I watched Governor Walz' press conference as it happened. He hit it
out of the park with his comments, in my opinion.
As an aside, the general officer in charge of the Minnesota National
Guard (who spoke at the press conference) is a personal friend of my
best friend. They served together for over a year, my friend was a
Lt. Colonel and Shawn Manke was a Major at the time. Per my friend,
Major General Manke is one of the good guys. But he will be put in
an unenviable position if Trump decides to federalize the MN Guard
Peter Fisher wrote:
Are you UKRM geeks who are still gainfully employed keeping an eye on
what the latest LLMs can do in the planning and coding arena?
If not, heads up, you need to.
I built my GPS speedo using Copilot. My Python skills are as rusty as
fuck, but it's done the job.
It still made loads of mistakes though.
I found the trick to make sure that Copilot didn't make mistakes, and
that was to repeat everything, including full dumps of the programmes
used, so that it corrected things properly.
I did idly wonder yesterday what would be best to say in the event
of a conversation with ICE.
Perhaps the toads don't know about Usenet. Let's hope not.
I built my GPS speedo using Copilot. [...] It still made loads of
mistakes though.
It's always like that: first you make it work then you improve--
it. Well, a lot of times, the process stops when it works. I don't
remember who said that code is not finish when there is nothing more
to add but when there is nothing more to remove.
That's why it's stupid to consider the best programmer as the one
who produce more lines of code than others.
And, from what I saw, actually, the AI produce a lot of code which
must be removed.
And, from what I saw, actually, the AI produce a lot of code which
must be removed.
The President said, "We simply cannot condone violence as a way of
changing the system ... Mob brutality, the total loss of respect for
human life was sickeningly sad ... What we saw last night and the
night before in Los Angeles is not about civil rights. It's not about
the great cause of equality that all Americans must uphold. It's not a >message of protest. It's been the brutality of a mob, pure and simple."
I'm also pretty uncomfortable sitting here wondering what to do after
the Gestapo/Brown Shirts v2.0 have just blatantly and transparently
murdered another person about 20 miles from me.
As an aside, the general officer in charge of the Minnesota National
Guard (who spoke at the press conference) is a personal friend of my
best friend. They served together for over a year, my friend was a
Lt. Colonel and Shawn Manke was a Major at the time. Per my friend,
Major General Manke is one of the good guys. But he will be put in
an unenviable position if Trump decides to federalize the MN Guard.
For complete transparency, I'm moving everything over to Proton
as I like the end to end encryption, the clients are intuitive to use,
and the two-person family plan just fits my needs.
Have you been reading my twitter (always will be that not X, for me) posts?
Hypocrisy sometimes goes beyond sickening.
Proton
On Sat, 24 Jan 2026 17:54:09 -0000 (UTC), Mark Olson
<olsonm@tiny.invalid> wrote:
I'm also pretty uncomfortable sitting here wondering what to do after
the Gestapo/Brown Shirts v2.0 have just blatantly and transparently >>murdered another person about 20 miles from me.
That's a different matter! Watching from this side of the pond, it is
both horrifying and scarcely believable
Those Minnosotans (sp?) turning out to protest, track ICE, support
their neighbourhoods, etc, are inspiring
On Sat, 24 Jan 2026 22:02:13 -0000 (UTC), Mark Olson
<olsonm@tiny.invalid> wrote:
As an aside, the general officer in charge of the Minnesota National
Guard (who spoke at the press conference) is a personal friend of my
best friend. They served together for over a year, my friend was a
Lt. Colonel and Shawn Manke was a Major at the time. Per my friend,
Major General Manke is one of the good guys. But he will be put in
an unenviable position if Trump decides to federalize the MN Guard.
I read that the MN Guard had turned out, wearing dayglo vests to
distinguish them from ICE, and were providing protestors with warm
drinks & food
https://www.msn.com/en-in/news/world/minnesota-guard-hands-out-coffee-to-anti-ice-protesters-after-deadly-federal-shooting/ar-AA1UXW1p
Looks like they've already picked a side
On Sat, 24 Jan 2026 18:32:36 +0000, Neil Ronketti
<neil@ronketti.invalid> wrote:
For complete transparency, I'm moving everything over to Proton
as I like the end to end encryption, the clients are intuitive to use,
and the two-person family plan just fits my needs.
<quick look at https://proton.me>
Ooh, interesting
Will look into this further
On Sat, 24 Jan 2026 18:36:07 +0000, Peter Fisher <nospam@nosspam.net>
wrote:
Have you been reading my twitter (always will be that not X, for me) posts?
You're still on twitter? Why?
On 26/01/2026 18:27, Champ wrote:
On Sat, 24 Jan 2026 18:32:36 +0000, Neil Ronketti
<neil@ronketti.invalid> wrote:
For complete transparency, I'm moving everything over to Proton
as I like the end to end encryption, the clients are intuitive to use,
and the two-person family plan just fits my needs.
<quick look at https://proton.me>
Ooh, interesting
Will look into this further
If you like what you see, I've got a referral code that gets anyone interested something like re415 off the subscription cost.
On 26/01/2026 22:11, Neil Ronketti wrote:
On 26/01/2026 18:27, Champ wrote:
On Sat, 24 Jan 2026 18:32:36 +0000, Neil Ronketti
<neil@ronketti.invalid> wrote:
For complete transparency, I'm moving everything over to Proton
as I like the end to end encryption, the clients are intuitive to use, >>>> and the two-person family plan just fits my needs.
<quick look at https://proton.me>
Ooh, interesting
Will look into this further
If you like what you see, I've got a referral code that gets anyone
interested something like re415 off the subscription cost.
I use the free tier as an 'extra' email address occasionally when
wanting to keep at arm's length, also their VPN if in foreign parts to
watch iPlayer.
Clearly I need to be more paranoid, but "life's too short..."
Are you UKRM geeks who are still gainfully employed keeping an eye on
what the latest LLMs can do in the planning and coding arena?
If not, heads up, you need to.
Champ <neal@champ.org.uk> wrote:
Those Minnosotans (sp?) turning out to protest, track ICE, support
their neighbourhoods, etc, are inspiring
I am humbled by what my neighbors have been up to in that way.
We did a couple of donations yesterday (food & other things) to support
folks who are afraid to leave home for fear of being kidnapped or worse.
My friend from Reykjavik emailed me this morning to say how upset
he is with what is happening to Minnesota. He offered to look for a
cottage for us in Iceland if we need to leave the US. He also said
he thinks his visit to the US in 2023 is probably his last.
On Sat, 24 Jan 2026 18:32:36 +0000, Neil Ronketti
<neil@ronketti.invalid> wrote:
For complete transparency, I'm moving everything over to Proton
as I like the end to end encryption, the clients are intuitive to use,
and the two-person family plan just fits my needs.
<quick look at https://proton.me>
Ooh, interesting
Will look into this further
You're still on twitter? Why?
I notice you haven't deleted your account.
Let's hope it won't come to it, but I think there's several people
over here who would take you in if you really need to become refugees
On Sat, 24 Jan 2026 18:32:36 +0000, Neil Ronketti
<neil@ronketti.invalid> wrote:
For complete transparency, I'm moving everything over to Proton
as I like the end to end encryption, the clients are intuitive to use,
and the two-person family plan just fits my needs.
<quick look at https://proton.me>
Ooh, interesting
Will look into this further
On Fri, 16 Jan 2026 08:20:55 +0000, Peter Fisher <nospam@nosspam.net>
wrote:
Are you UKRM geeks who are still gainfully employed keeping an eye on
what the latest LLMs can do in the planning and coding arena?
If not, heads up, you need to.
And another El Reg article on the usefulness of AI in software
development
https://www.theregister.com/2026/01/26/cursor_opinion/
On 27/01/2026 10:21, Champ wrote:
Let's hope it won't come to it, but I think there's several people
over here who would take you in if you really need to become refugees
Anyone in Minnesota (or the rest of the USA come to think if it)
probably has a legitimate case for seeking asylum IWHT.
On 27/01/2026 08:15, Champ wrote:
On Fri, 16 Jan 2026 08:20:55 +0000, Peter Fisher <nospam@nosspam.net>
wrote:
Are you UKRM geeks who are still gainfully employed keeping an eye on
what the latest LLMs can do in the planning and coding arena?
If not, heads up, you need to.
And another El Reg article on the usefulness of AI in software
development
https://www.theregister.com/2026/01/26/cursor_opinion/
Sure. ".., but marketing keeps insisting junior engineers can take whole projects from spec to shipping." Anyone who has tried a decent AI tuned
for code development knows this is this not true, however, I don't think that means AI coding can be dismissed out of hand.
They are also useful to work out why a system that used to work, crashes
- in particular one on a VPS after routine maintenance by the hosting platform meant it was shutdown. Mostly rebooted OK, but my Java based
DMS turned its toes up. Reinstall from scratch or backup could have
worked, but Claude pointed me to the logs that would tell the tale and
how to resolve the issue. Then opined that it could run much faster (the 'server'-a right on the spec limit) and suggested tweaks, which have definitely improved it. They might not work if it has getting 1000s of requests a day TBF - but my 'user base' is tiny.
On 27/01/2026 12:12, Simon Wilson wrote:
On 27/01/2026 10:21, Champ wrote:
Let's hope it won't come to it, but I think there's several people
over here who would take you in if you really need to become refugees
Anyone in Minnesota (or the rest of the USA come to think if it)
probably has a legitimate case for seeking asylum IWHT.
Indeed.
Not sure if this is positive news or not: https://www.theguardian.com/us-news/2026/jan/27/gregory-bovino-minneapolis-minnesota-alex-pretti-shooting
Peter Fisher <nospam@nosspam.net> wrote:
On 27/01/2026 12:12, Simon Wilson wrote:
On 27/01/2026 10:21, Champ wrote:
Let's hope it won't come to it, but I think there's several people
over here who would take you in if you really need to become refugees
Anyone in Minnesota (or the rest of the USA come to think if it)
probably has a legitimate case for seeking asylum IWHT.
Indeed.
Not sure if this is positive news or not:
https://www.theguardian.com/us-news/2026/jan/27/gregory-bovino-minneapolis-minnesota-alex-pretti-shooting
I sincerely doubt it.
https://www.reddit.com/r/minnesota/comments/1qoiep8/homan_contributed_to_project_2025s_immigration/
On 27/01/2026 18:09, Mark Olson wrote:
Peter Fisher <nospam@nosspam.net> wrote:
Not sure if this is positive news or not:
https://www.theguardian.com/us-news/2026/jan/27/gregory-bovino-minneapolis-minnesota-alex-pretti-shooting
I sincerely doubt it.
https://www.reddit.com/r/minnesota/comments/1qoiep8/homan_contributed_to_project_2025s_immigration/
Oh, I see now, the organ grinder's monkey being replaced by a higher
status monkey. :(
It must be really scary. I'm sorry. Seems there's very little I can do
other than try to avoid buying anything made in the USA. Or travelling there.
I'm also pretty uncomfortable sitting here wondering what to do after
the Gestapo/Brown Shirts v2.0 have just blatantly and transparently
murdered another person about 20 miles from me.
On 25/1/26 00:54, Mark Olson wrote:
I'm also pretty uncomfortable sitting here wondering what to do after
the Gestapo/Brown Shirts v2.0 have just blatantly and transparently >>murdered another person about 20 miles from me.
From my statistics mate in Alaska:
"Little known fact: There are ten times more Gundersons per capita in Minnesota than there are in the rest of the United States. And Olsons and Nygaards too."
You'll be OK, sounds like you are surrounded by plenty of good people.
As mentioned elsewhere, if the shit really hits the fan, your asylum application here would be nothing more than a formality.
Now, about the trip to Mnnchen and Ljubljana. Only problem there is trying
to pronounce the latter without getting one's cheeks caught between one's back teeth.
Let me know when you want to talk beer.
"IDEs/agent swarms/fallability. Both the "no need for IDE anymore" hype
and the "agent swarm" hype is imo too much for right now. The models definitely still make mistakes and if you have any code you actually
care about I would watch them like a hawk, in a nice large IDE on the
side. The mistakes have changed a lot - they are not simple syntax
errors anymore, they are subtle conceptual errors that a slightly
sloppy, hasty junior dev might do.
On 27/01/2026 16:22, Peter Fisher wrote:
"IDEs/agent swarms/fallability. Both the "no need for IDE anymore"
hype and the "agent swarm" hype is imo too much for right now. The
models definitely still make mistakes and if you have any code you
actually care about I would watch them like a hawk, in a nice large
IDE on the side. The mistakes have changed a lot - they are not simple
syntax errors anymore, they are subtle conceptual errors that a
slightly sloppy, hasty junior dev might do.
<SNIP>
I've started asking people at work to think of AI exactly as they would
a sixth-former on work experience who's been assigned to them. Don't be amazed that they can write in complete sentences, or deliver lines of
code. Instead, check every output with exactly the same thoroughness you would if the spotty youth had produced it.
I also invite them to consider if they really think their job would be better if it was reduced to explaining, repeatedly, what they wanted the work-experience kid to do, and then checking it was done right.
Finally, don't ask it to produce something that you can't or won't
check, and if you can't be bothered to check it then don't assume anyone else will.
It infuriates me when, to pick an example, I ask our (very non-
technical) head of marketing what she wants to use our CRM system for,
and get back 4 pages of step-by-step instructions on how to set it up
and integrate it with 3rd party systems. She hasn't checked that it's correct. I can't check that it's correct, because she still hasn't told
me what she's trying to do. Nobody knows if the third-party systems it mentions exist or do what the AI thinks they do. AI has boosted
productivity by saving her an hour of organising her thoughts into a
couple of paragraphs of English, for the low cost of sending us down a
road that involves paying for other services, possibly needlessly, to
get a solution that might, for all we know, just email dick-pics to all
our customers.
On Mon, 26 Jan 2026 22:33:43 +0000, Peter Fisher <nospam@nosspam.net>
wrote:
You're still on twitter? Why?
I notice you haven't deleted your account.
I thought I had. But then, with the state of that system now, it
probably took me through the steps of deleting it, and left stuff
there for other people to see
I'll look at it again
I also invite them to consider if they really think their job would be >better if it was reduced to explaining, repeatedly, what they wanted the >work-experience kid to do, and then checking it was done right.
On Wed, 28 Jan 2026 08:58:51 +0000, AndrewR
<andrew.rockface.spam@gmail.com> wrote:
<of AI>
I also invite them to consider if they really think their job would be
better if it was reduced to explaining, repeatedly, what they wanted the
work-experience kid to do, and then checking it was done right.
Exactly. If the dev job becomes supervising and correcting the
machines, that, imo, is a serious downgrade
On Tue, 27 Jan 2026 10:19:23 +0000, Champ <neal@champ.org.uk> wrote:
On Mon, 26 Jan 2026 22:33:43 +0000, Peter Fisher <nospam@nosspam.net>
wrote:
You're still on twitter? Why?
I notice you haven't deleted your account.
I thought I had. But then, with the state of that system now, it
probably took me through the steps of deleting it, and left stuff
there for other people to see
I'll look at it again
Done it now, although apparently it still lingers for 30 days
I'm not go to hang our in a place full of fascists and which is used
to generate pornography (inc of children)
Champ wrote:
On Wed, 28 Jan 2026 08:58:51 +0000, AndrewR
<andrew.rockface.spam@gmail.com> wrote:
<of AI>
I also invite them to consider if they really think their job would
be better if it was reduced to explaining, repeatedly, what they
wanted the work-experience kid to do, and then checking it was done
right.
Exactly.-a If the dev job becomes supervising and correcting the
machines, that, imo, is a serious downgrade
Here is a nice article by Cory Doctorow explaining exactly why all the
AI investment is intended to achieve exactly this (with the side effect
of propping up the share prices of the AI vendors, regardless of whether they are successful). There is also the added 'benefit' that the human supervisor also takes the blame for the AI's mistakes.
https://www.theguardian.com/us-news/ng-interactive/2026/jan/18/tech-ai- bubble-burst-reverse-centaur
So where's the cool echo chamber place to hang out with the right sort
of people these days - Mastodon or Bluesky? I still subscribe to DAG's substack.
So where's the cool echo chamber place to hang out with the right sort
of people these days - Mastodon or Bluesky? I still subscribe to DAG's substack.
On Tue, 27 Jan 2026 18:05:59 -0000 (UTC), Mark Olson
<olsonm@tiny.invalid> wrote:
Champ <neal@champ.org.uk> wrote:
Let's hope it won't come to it, but I think there's several people
over here who would take you in if you really need to become refugees
Honestly, that made me tear up a bit.
Thanks.
No problem, and (from my pov) it's a genuine offer.
Of course, from your pov, I don't underestimate the wrench that
relocating across the world because your own nation is no longer a
place you want to live would be!
On 24/01/2026 18:08, Simon Wilson wrote:
It must be really scary. I'm sorry. Seems there's very little I can do
other than try to avoid buying anything made in the USA. Or travelling
there.
Due to having to communicate with other people, I can't avoid using FB, Whatsapp, and certain aspects of Google. My interactions with Amazon are minimal (my b-i-l sends me an Amazon card for Christmas each year, and that's about it). My main expenditure is on musical instruments and amplification - I have one American made instrument, won't be buying any more, and I did have one homebuild cab with an (expensive) American
Eminence speaker in it which died horribly, demonstrating the inferior quality of "Made in USA" things. It's being replaced with a proper Made
in China speaker.
Peter Fisher <nospam@nosspam.net> wrote:
So where's the cool echo chamber place to hang out with the right sort
of people these days - Mastodon or Bluesky? I still subscribe to DAG's
substack.
Mastodon. Bluesky suffers from the same centralized control problem
as Twitter, as I understand it.
There's a goodly number of UKRMers there and I'm following most of
them if I can figure out who they are.
Me, YTC#1, Timo, Eddie, and more.
I'll let someone who is more knowledgeable about which one of the
many instances (servers) you can/should sign up with chime in.
Are you UKRM geeks who are still gainfully employed keeping an eye on
what the latest LLMs can do in the planning and coding arena?
If not, heads up, you need to.
On 28/01/2026 14:56, Mark Olson wrote:
Peter Fisher <nospam@nosspam.net> wrote:
So where's the cool echo chamber place to hang out with the right sort
of people these days - Mastodon or Bluesky? I still subscribe to DAG's
substack.
Mastodon. Bluesky suffers from the same centralized control problem
as Twitter, as I understand it.
There's a goodly number of UKRMers there and I'm following most of
them if I can figure out who they are.
Me, YTC#1, Timo, Eddie, and more.
And me :-)
chrisnd @ukrm <chrisnd@privacy.net> wrote:
On 28/01/2026 14:56, Mark Olson wrote:
Peter Fisher <nospam@nosspam.net> wrote:
So where's the cool echo chamber place to hang out with the right sort >>>> of people these days - Mastodon or Bluesky? I still subscribe to DAG's >>>> substack.
Mastodon. Bluesky suffers from the same centralized control problem
as Twitter, as I understand it.
There's a goodly number of UKRMers there and I'm following most of
them if I can figure out who they are.
Me, YTC#1, Timo, Eddie, and more.
And me :-)
And Champ, and AndrewR, AstroSpanner, Steven Packer, etc.
I'm elkster88@ ... I'm not sure if I am following you or not!
NB: No, the 88 isn't some secret Nazi code :( as I have just now
realized some folk might think. Entirely a coincidence but it's too
late now to undo it.
Peter Fisher <nospam@nosspam.net> wrote:
So where's the cool echo chamber place to hang out with the right sort
of people these days - Mastodon or Bluesky? I still subscribe to DAG's
substack.
Mastodon. Bluesky suffers from the same centralized control problem
as Twitter, as I understand it.
There's a goodly number of UKRMers there and I'm following most of
them if I can figure out who they are.
Me, YTC#1, Timo, Eddie, and more.
I'll let someone who is more knowledgeable about which one of the
many instances (servers) you can/should sign up with chime in.
Peter Fisher <nospam@nosspam.net> wrote in news:10kcsd5$1eute$1@dont- email.me:
Are you UKRM geeks who are still gainfully employed keeping an eye on
what the latest LLMs can do in the planning and coding arena?
If not, heads up, you need to.
https://www.bbc.co.uk/sounds/play/m002qj6b
Aleks and Kevin explore the biggest story in the Ai business. Markets are growing sceptical that the eye-watering sums being invested in Ai will show
a return and comparisons with the dot-com crash are being widely made - so
is the Ai bubble about to burst?
Over the next three episodes we'll explore the Ai business landscape, look
at the evidence for a bubble and consider the consequences of a 'major correction' to the valuation of the big players.
Presenters: Aleks Krotoski and Kevin Fo
Peter Fisher wrote:
So where's the cool echo chamber place to hang out with the right sort
of people these days - Mastodon or Bluesky? I still subscribe to DAG's
substack.
Philip Bump, ex-Washington Post columnist -
"Substack, a company that intentionally provides a voice to white nationalist and Nazi propaganda." https://www.pbump.net/o/the-power-we-use-and-the-power-we-give/
Unfortunately, anywhere owned by somebody else - particularly 'tech-bro' billionaires (Twitter, Meta), venture capitalists (TikTok), crypto-
grifters (BlueSky) etc. is going to be a problem, whether they have an agenda to push or just want more (and more) billions.
Google and Apple don't care, so long as they get their 30% cut - https://www.theregister.com/2026/01/27/ nudify_app_proliferation_shows_naked/
Open-source and federated (eg. Usenet or ActivityPub based platforms
like Mastodon, Lemmy etc.) or independently owned fora are really the
only way.
NB: No, the 88 isn't some secret Nazi code :( as I have just now
realized some folk might think. Entirely a coincidence but it's too
late now to undo it.
chrisnd @ukrm <chrisnd@privacy.net> wrote:
On 28/01/2026 14:56, Mark Olson wrote:
Peter Fisher <nospam@nosspam.net> wrote:
So where's the cool echo chamber place to hang out with the right sort >>>> of people these days - Mastodon or Bluesky? I still subscribe to DAG's >>>> substack.
Mastodon. Bluesky suffers from the same centralized control problem
as Twitter, as I understand it.
There's a goodly number of UKRMers there and I'm following most of
them if I can figure out who they are.
Me, YTC#1, Timo, Eddie, and more.
And me :-)
And Champ, and AndrewR, AstroSpanner, Steven Packer, etc.
I'm elkster88@ ... I'm not sure if I am following you or not!
NB: No, the 88 isn't some secret Nazi code :( as I have just now
realized some folk might think. Entirely a coincidence but it's too
late now to undo it.
On Jan 28, 2026 at 11:30:24 AM EST, "Mark Olson" <olsonm@tiny.invalid> wrote:
NB: No, the 88 isn't some secret Nazi code :( as I have just now
realized some folk might think. Entirely a coincidence but it's too
late now to undo it.
88 in British culture is synomous with "Two Fat Ladies" - a bingo call. People will probably just think you're a chubby-chaser.[0]
[0] Not that there's anything wrong with that.
Just logged in. Am I missing something, or is there not a lot of
'original content' posted by UKRMers? ISTR I have to add new peeps to
some kind of list to see them gathered in the Church of Reason that is
the UKRM contingent.
Ben Blaney <benblaney@gmail.invalid> wrote:
On Jan 28, 2026 at 11:30:24 AM EST, "Mark Olson" <olsonm@tiny.invalid> wrote:
NB: No, the 88 isn't some secret Nazi code :( as I have just now
realized some folk might think. Entirely a coincidence but it's too
late now to undo it.
88 in British culture is synomous with "Two Fat Ladies" - a bingo call.
People will probably just think you're a chubby-chaser.[0]
[0] Not that there's anything wrong with that.
Never heard that one up till now. I'd be happy to be mistaken for
one of them, without shame!
It is beyond doubt that Trump has already crossed the
Rubicon/Mississippi/St. Croix.
On Fri, 16 Jan 2026 08:20:55 +0000, Peter Fisher <nospam@nosspam.net>
wrote:
Are you UKRM geeks who are still gainfully employed keeping an eye on
what the latest LLMs can do in the planning and coding arena?
If not, heads up, you need to.
And another El Reg article on the usefulness of AI in software
development
https://www.theregister.com/2026/01/26/cursor_opinion/
On Tue, 27 Jan 2026 08:15:52 +0000, Champ <neal@champ.org.uk> wrote:
On Fri, 16 Jan 2026 08:20:55 +0000, Peter Fisher <nospam@nosspam.net>
wrote:
Are you UKRM geeks who are still gainfully employed keeping an eye on
what the latest LLMs can do in the planning and coding arena?
If not, heads up, you need to.
And another El Reg article on the usefulness of AI in software
development
https://www.theregister.com/2026/01/26/cursor_opinion/
...and another one: https://www.theregister.com/2026/01/28/claude_code_ai_secrets_files/
https://www.bbc.co.uk/sounds/play/m002qj6b
Aleks and Kevin [] look at the evidence for a bubble
On 28/01/2026 18:59, Mark Olson wrote:
Ben Blaney <benblaney@gmail.invalid> wrote:
On Jan 28, 2026 at 11:30:24 AM EST, "Mark Olson"
<olsonm@tiny.invalid> wrote:
NB: No, the 88 isn't some secret Nazi code :( as I have just now
realized some folk might think. Entirely a coincidence but it's too
late now to undo it.
88 in British culture is synomous with "Two Fat Ladies" - a bingo call.
People will probably just think you're a chubby-chaser.[0]
[0]-a Not that there's anything wrong with that.
Never heard that one up till now.-a I'd be happy to be mistaken for
one of them, without shame!
"Picked her up in my 88."
Two fat ladies has a particular connection with a UK TV cookery series. https://www.bbc.co.uk/iplayer/episode/b008l31m/two-fat-ladies-christmas
Both long gone. A liking for butter, wine and fags may have contributed
to their demise. Nice sidecar outfit featured.
Peter Fisher <nospam@nosspam.net> wrote:
Just logged in. Am I missing something, or is there not a lot of
'original content' posted by UKRMers? ISTR I have to add new peeps to
some kind of list to see them gathered in the Church of Reason that is
the UKRM contingent.
I mostly only see favorites, boosts and "toots" from Eddie and YTC#1,
but I'm following most of the UKRMers that I could identify. I don't
have any lists defined. I think most of our crowd simply aren't all
that active, BICBW.
chrisnd @ukrm <chrisnd@privacy.net> wrote:
On 28/01/2026 14:56, Mark Olson wrote:
Peter Fisher <nospam@nosspam.net> wrote:
Mastodon. Bluesky suffers from the same centralized control problem
as Twitter, as I understand it.
There's a goodly number of UKRMers there and I'm following most of
them if I can figure out who they are.
Me, YTC#1, Timo, Eddie, and more.
And me :-)
And Champ, and AndrewR, AstroSpanner, Steven Packer, etc.
On 28/01/2026 16:30, Mark Olson wrote:
chrisnd @ukrm <chrisnd@privacy.net> wrote:
On 28/01/2026 14:56, Mark Olson wrote:
Peter Fisher <nospam@nosspam.net> wrote:
Mastodon.-a Bluesky suffers from the same centralized control problem
as Twitter, as I understand it.
There's a goodly number of UKRMers there and I'm following most of
them if I can figure out who they are.
Me, YTC#1, Timo, Eddie, and more.
And me :-)
And Champ, and AndrewR, AstroSpanner, Steven Packer, etc.
I haven't posted anything on Mastodon in ages. Last time I checked in I
had a notification telling me that the admins had defederated a server,
so I could no longer follow 5 of the people I'd been following... and
I'd given up on it because I'd predicted that sort of thing was going to happen.
Centralised control might be a problem but de-centralised control means thousands of local admins, answerable to no-one, and they'll all be cunts.
Peter Fisher <nospam@nosspam.net> wrote:<waves>
Just logged in. Am I missing something, or is there not a lot of
'original content' posted by UKRMers? ISTR I have to add new peeps to
some kind of list to see them gathered in the Church of Reason that is
the UKRM contingent.
I mostly only see favorites, boosts and "toots" from Eddie and YTC#1,
but I'm following most of the UKRMers that I could identify. I don'tI like the old school techie stuff on there
have any lists defined. I think most of our crowd simply aren't allAge, does for the joints
that active, BICBW.
On 28/01/2026 16:30, Mark Olson wrote:
chrisnd @ukrm <chrisnd@privacy.net> wrote:
On 28/01/2026 14:56, Mark Olson wrote:
Peter Fisher <nospam@nosspam.net> wrote:
Mastodon.-a Bluesky suffers from the same centralized control problem
as Twitter, as I understand it.
There's a goodly number of UKRMers there and I'm following most of
them if I can figure out who they are.
Me, YTC#1, Timo, Eddie, and more.
And me :-)
And Champ, and AndrewR, AstroSpanner, Steven Packer, etc.
I haven't posted anything on Mastodon in ages. Last time I checked in I
had a notification telling me that the admins had defederated a server,
so I could no longer follow 5 of the people I'd been following... and
I'd given up on it because I'd predicted that sort of thing was going to happen.
Centralised control might be a problem but de-centralised control means thousands of local admins, answerable to no-one, and they'll all be cunts.
On 29/01/2026 15:49, AndrewR wrote:
On 28/01/2026 16:30, Mark Olson wrote:Ah indeed.-a Federation is a great idea but susceptible to this kind of problem.-a Unfortunately.
chrisnd @ukrm <chrisnd@privacy.net> wrote:
On 28/01/2026 14:56, Mark Olson wrote:
Peter Fisher <nospam@nosspam.net> wrote:
Mastodon.-a Bluesky suffers from the same centralized control problem >>>>> as Twitter, as I understand it.
There's a goodly number of UKRMers there and I'm following most of
them if I can figure out who they are.
Me, YTC#1, Timo, Eddie, and more.
And me :-)
And Champ, and AndrewR, AstroSpanner, Steven Packer, etc.
I haven't posted anything on Mastodon in ages. Last time I checked in
I had a notification telling me that the admins had defederated a
server, so I could no longer follow 5 of the people I'd been
following... and I'd given up on it because I'd predicted that sort of
thing was going to happen.
Centralised control might be a problem but de-centralised control
means thousands of local admins, answerable to no-one, and they'll all
be cunts.
Unfortunately also, websites are similarly susceptible.-a Locally (in a literal sense) a former local history website disappeared in the early flowerings of the internet, following the (unannounced) death of its
owner, before I had chance to download a hitherto unknown photo of my grandmother as a school girl.
On 29/01/2026 16:42, chrisnd @ukrm wrote:
On 29/01/2026 15:49, AndrewR wrote:
On 28/01/2026 16:30, Mark Olson wrote:Ah indeed.-a Federation is a great idea but susceptible to this kind of
chrisnd @ukrm <chrisnd@privacy.net> wrote:
On 28/01/2026 14:56, Mark Olson wrote:
Peter Fisher <nospam@nosspam.net> wrote:
Mastodon.-a Bluesky suffers from the same centralized control problem >>>>>> as Twitter, as I understand it.
There's a goodly number of UKRMers there and I'm following most of >>>>>> them if I can figure out who they are.
Me, YTC#1, Timo, Eddie, and more.
And me :-)
And Champ, and AndrewR, AstroSpanner, Steven Packer, etc.
I haven't posted anything on Mastodon in ages. Last time I checked in
I had a notification telling me that the admins had defederated a
server, so I could no longer follow 5 of the people I'd been
following... and I'd given up on it because I'd predicted that sort
of thing was going to happen.
Centralised control might be a problem but de-centralised control
means thousands of local admins, answerable to no-one, and they'll
all be cunts.
problem.-a Unfortunately.
Unfortunately also, websites are similarly susceptible.-a Locally (in a
literal sense) a former local history website disappeared in the early
flowerings of the internet, following the (unannounced) death of its
owner, before I had chance to download a hitherto unknown photo of my
grandmother as a school girl.
Pervert
On 28/01/2026 16:12, wessie wrote:
Peter Fisher <nospam@nosspam.net> wrote in news:10kcsd5$1eute$1@dont-
email.me:
Are you UKRM geeks who are still gainfully employed keeping an eye on
what the latest LLMs can do in the planning and coding arena?
If not, heads up, you need to.
https://www.bbc.co.uk/sounds/play/m002qj6b
Aleks and Kevin explore the biggest story in the Ai business. Markets are
growing sceptical that the eye-watering sums being invested in Ai will show >> a return and comparisons with the dot-com crash are being widely made - so >> is the Ai bubble about to burst?
Over the next three episodes we'll explore the Ai business landscape, look >> at the evidence for a bubble and consider the consequences of a 'major
correction' to the valuation of the big players.
Presenters: Aleks Krotoski and Kevin Fo
Sun was shining, so I was out in the garden dealing with pond sludge (literally, not metaphorically), but I heard the trailer and will listen
to the podcast.
Peter Fisher <nospam@nosspam.net> wrote:
On 28/01/2026 16:12, wessie wrote:
Peter Fisher <nospam@nosspam.net> wrote in
news:10kcsd5$1eute$1@dont- email.me:
Are you UKRM geeks who are still gainfully employed keeping an eye
on what the latest LLMs can do in the planning and coding arena?
If not, heads up, you need to.
https://www.bbc.co.uk/sounds/play/m002qj6b
Aleks and Kevin explore the biggest story in the Ai business.
Markets are growing sceptical that the eye-watering sums being
invested in Ai will show a return and comparisons with the dot-com
crash are being widely made - so is the Ai bubble about to burst?
Over the next three episodes we'll explore the Ai business
landscape, look at the evidence for a bubble and consider the
consequences of a 'major correction' to the valuation of the big
players.
Presenters: Aleks Krotoski and Kevin Fo
Sun was shining, so I was out in the garden dealing with pond sludge
(literally, not metaphorically), but I heard the trailer and will
listen to the podcast.
That was worth listening to. Somewhere between apocalypse any time
soon and quite useful in niche areas like coding.
Aren't all admins cunts?
On 29/01/2026 16:50, YTC1 wrote:
On 29/01/2026 16:42, chrisnd @ukrm wrote:
On 29/01/2026 15:49, AndrewR wrote:
On 28/01/2026 16:30, Mark Olson wrote:Ah indeed.-a Federation is a great idea but susceptible to this kind
chrisnd @ukrm <chrisnd@privacy.net> wrote:
On 28/01/2026 14:56, Mark Olson wrote:
Peter Fisher <nospam@nosspam.net> wrote:
Mastodon.-a Bluesky suffers from the same centralized control problem >>>>>>> as Twitter, as I understand it.
There's a goodly number of UKRMers there and I'm following most of >>>>>>> them if I can figure out who they are.
Me, YTC#1, Timo, Eddie, and more.
And me :-)
And Champ, and AndrewR, AstroSpanner, Steven Packer, etc.
I haven't posted anything on Mastodon in ages. Last time I checked
in I had a notification telling me that the admins had defederated a
server, so I could no longer follow 5 of the people I'd been
following... and I'd given up on it because I'd predicted that sort
of thing was going to happen.
Centralised control might be a problem but de-centralised control
means thousands of local admins, answerable to no-one, and they'll
all be cunts.
of problem.-a Unfortunately.
Unfortunately also, websites are similarly susceptible.-a Locally (in
a literal sense) a former local history website disappeared in the
early flowerings of the internet, following the (unannounced) death
of its owner, before I had chance to download a hitherto unknown
photo of my grandmother as a school girl.
Pervert
I was just checking the old spirit of UKRM was still alive in some
places :-)
However, I was disappointed the other day when a 'ginge' opportunity was missed...
Although the perpetrator did seem to ginge themselves?-a Make of that
what you will!
Pete Fisher <no_email@invalid.invalid> wrote in news:10lge8g$1iqvd$1@dont-email.me:
Peter Fisher <nospam@nosspam.net> wrote:
On 28/01/2026 16:12, wessie wrote:
Peter Fisher <nospam@nosspam.net> wrote in
news:10kcsd5$1eute$1@dont- email.me:
Are you UKRM geeks who are still gainfully employed keeping an eye
on what the latest LLMs can do in the planning and coding arena?
If not, heads up, you need to.
https://www.bbc.co.uk/sounds/play/m002qj6b
Aleks and Kevin explore the biggest story in the Ai business.
Markets are growing sceptical that the eye-watering sums being
invested in Ai will show a return and comparisons with the dot-com
crash are being widely made - so is the Ai bubble about to burst?
Over the next three episodes we'll explore the Ai business
landscape, look at the evidence for a bubble and consider the
consequences of a 'major correction' to the valuation of the big
players.
Presenters: Aleks Krotoski and Kevin Fo
Sun was shining, so I was out in the garden dealing with pond sludge
(literally, not metaphorically), but I heard the trailer and will
listen to the podcast.
That was worth listening to. Somewhere between apocalypse any time
soon and quite useful in niche areas like coding.
did you listen to the other 2 episodes already on Sounds? I have not, yet, as I caught it on linear R4
On 29/01/2026 17:09, chrisnd @ukrm wrote:
On 29/01/2026 16:50, YTC1 wrote:
On 29/01/2026 16:42, chrisnd @ukrm wrote:
On 29/01/2026 15:49, AndrewR wrote:
On 28/01/2026 16:30, Mark Olson wrote:Ah indeed.-a Federation is a great idea but susceptible to this kind
chrisnd @ukrm <chrisnd@privacy.net> wrote:
On 28/01/2026 14:56, Mark Olson wrote:
Peter Fisher <nospam@nosspam.net> wrote:
Mastodon.-a Bluesky suffers from the same centralized control >>>>>>>> problem
as Twitter, as I understand it.
There's a goodly number of UKRMers there and I'm following most of >>>>>>>> them if I can figure out who they are.
Me, YTC#1, Timo, Eddie, and more.
And me :-)
And Champ, and AndrewR, AstroSpanner, Steven Packer, etc.
I haven't posted anything on Mastodon in ages. Last time I checked
in I had a notification telling me that the admins had defederated
a server, so I could no longer follow 5 of the people I'd been
following... and I'd given up on it because I'd predicted that sort >>>>> of thing was going to happen.
Centralised control might be a problem but de-centralised control
means thousands of local admins, answerable to no-one, and they'll
all be cunts.
of problem.-a Unfortunately.
Unfortunately also, websites are similarly susceptible.-a Locally (in >>>> a literal sense) a former local history website disappeared in the
early flowerings of the internet, following the (unannounced) death
of its owner, before I had chance to download a hitherto unknown
photo of my grandmother as a school girl.
Pervert
I was just checking the old spirit of UKRM was still alive in some
places :-)
However, I was disappointed the other day when a 'ginge' opportunity
was missed...
Although the perpetrator did seem to ginge themselves?-a Make of that
what you will!
Did I miss that while I was in Greece?
So where's the cool echo chamber place to hang out with the right sort
of people these days - Mastodon or Bluesky?
On Wed, 28 Jan 2026 13:06:19 +0000, Peter Fisher <nospam@nosspam.net>
wrote:
So where's the cool echo chamber place to hang out with the right sort
of people these days - Mastodon or Bluesky?
Bluesky for me. It has the critical mass to make it worthwhile, and
all the people I want to follow are on there [1]
[1] Politics: Ian Dunt Dorian Lyndsky, David Allen Green
Bike Racing: David Emmett, Matt Oxley, Simon Patterson
On 31/01/2026 09:49, Champ wrote:
On Wed, 28 Jan 2026 13:06:19 +0000, Peter Fisher <nospam@nosspam.net> >>wrote:
So where's the cool echo chamber place to hang out with the right sort >>>of people these days - Mastodon or Bluesky?
Bluesky for me. It has the critical mass to make it worthwhile, and
all the people I want to follow are on there [1]
[1] Politics: Ian Dunt Dorian Lyndsky, David Allen Green
Bike Racing: David Emmett, Matt Oxley, Simon Patterson
Will take a look. One or two people are whT keeps mein the hell site
Pete Fisher <spam@nospam.net> wrote:
On 31/01/2026 09:49, Champ wrote:
On Wed, 28 Jan 2026 13:06:19 +0000, Peter Fisher <nospam@nosspam.net> >>>wrote:
So where's the cool echo chamber place to hang out with the right sort >>>>of people these days - Mastodon or Bluesky?
Bluesky for me. It has the critical mass to make it worthwhile, and
all the people I want to follow are on there [1]
[1] Politics: Ian Dunt Dorian Lyndsky, David Allen Green
Bike Racing: David Emmett, Matt Oxley, Simon Patterson
Will take a look. One or two people are whT keeps mein the hell site
Subliminal typo?
On Wed, 28 Jan 2026 13:06:19 +0000, Peter Fisher <nospam@nosspam.net>
wrote:
So where's the cool echo chamber place to hang out with the right sort
of people these days - Mastodon or Bluesky?
Bluesky for me. It has the critical mass to make it worthwhile, and
all the people I want to follow are on there [1]
[1] Politics: Ian Dunt Dorian Lyndsky, David Allen Green
Bike Racing: David Emmett, Matt Oxley, Simon Patterson
On 31/01/2026 09:49, Champ wrote:
On Wed, 28 Jan 2026 13:06:19 +0000, Peter Fisher <nospam@nosspam.net>
wrote:
So where's the cool echo chamber place to hang out with the right sort
of people these days - Mastodon or Bluesky?
Bluesky for me. It has the critical mass to make it worthwhile, and
all the people I want to follow are on there [1]
I have to agree, and handily some people keep us informed at what is
being said on twitter.
I'd like more news[1] services and politicians to appear though.
[1] Politics: Ian Dunt Dorian Lyndsky, David Allen Green
Bike Racing: David Emmett, Matt Oxley, Simon Patterson
#1 Tomorrows Papers Today is handy though.
On Wed, 28 Jan 2026 13:06:19 +0000, Peter Fisher <nospam@nosspam.net>
wrote:
So where's the cool echo chamber place to hang out with the right sort
of people these days - Mastodon or Bluesky?
Bluesky for me. It has the critical mass to make it worthwhile, and
all the people I want to follow are on there [1]
[1] Politics: Ian Dunt Dorian Lyndsky, David Allen Green
Bike Racing: David Emmett, Matt Oxley, Simon Patterson
YTC1 <ytc1@ytc1.co.uk> wrote in news:10ln9k3$3kmj5$3@dont-email.me:No, as it is is just the paper front pages being shown. I can choose to
On 31/01/2026 09:49, Champ wrote:
On Wed, 28 Jan 2026 13:06:19 +0000, Peter Fisher <nospam@nosspam.net>
wrote:
So where's the cool echo chamber place to hang out with the right sort >>>> of people these days - Mastodon or Bluesky?
Bluesky for me. It has the critical mass to make it worthwhile, and
all the people I want to follow are on there [1]
I have to agree, and handily some people keep us informed at what is
being said on twitter.
I'd like more news[1] services and politicians to appear though.
[1] Politics: Ian Dunt Dorian Lyndsky, David Allen Green
Bike Racing: David Emmett, Matt Oxley, Simon Patterson
#1 Tomorrows Papers Today is handy though.
isn't that when it fills with slop from the likes of Reach plc [0] and lies from populists i.e. enshittfies
Bluesky for me. It has the critical mass to make it worthwhile, and
all the people I want to follow are on there [1]
[1] Politics: Ian Dunt Dorian Lyndsky, David Allen GreenI like Ian Dunt, but he writes about how he won't use Twitter because it
Bike Racing: David Emmett, Matt Oxley, Simon Patterson
Two fat ladies has a particular connection with a UK TV cookery series. https://www.bbc.co.uk/iplayer/episode/b008l31m/two-fat-ladies-christmas
Both long gone. A liking for butter, wine and fags may have contributed
to their demise. Nice sidecar outfit featured.
folks who are afraid to leave home for fear of being kidnapped or worse.
Champ wrote:
Bluesky for me. It has the critical mass to make it worthwhile, and
all the people I want to follow are on there [1]
[1] Politics: Ian Dunt Dorian Lyndsky, David Allen GreenI like Ian Dunt, but he writes about how he won't use Twitter because it
Bike Racing: David Emmett, Matt Oxley, Simon Patterson
is run and used by Nazis, then publishes on Substack - who allow and
profit from Nazi content and 'accidentally' push it to users: >https://arstechnica.com/tech-policy/2025/07/substacks-nazi-problem-wont-go-away-after-push-notification-apology/
Bluesky was founded by Jack Dorsey (also Twitter founder) who really,
really tried to prevent neo-Nazis being banned from Twitter.
He left Bluesky when they implemented moderation, which he really,
really did not want. >[https://davidgerard.co.uk/blockchain/2024/05/10/jack-dorsey-bluesky-decentralised-social-networks-and-the-very-common-crowd/]
Bluesky is funded by venture capital ("Blockchain Capital is a venture >capital firm focused on bitcoin and cryptocurrency-related projects").
It promises federation, but hasn't delivered (and won't).
It is a textbook case of the first stage of enshitification (play nice,
lock in users due to network effect); the rest of the enshitification
cycle will surely follow.
I am humbled by what my neighbors have been up to in that way.
On Jan 26, 2026 at 2:31:06 PM EST, "Mark Olson" <olsonm@tiny.invalid> wrote:
I am humbled by what my neighbors have been up to in that way.
You might be interested to know that Billy Bragg has written a protest song, inspired my friend Steve Rapport's reaction to Minneapolis.
https://consequence.net/2026/02/billy-bragg-minneapolis-guest-essay-snowy-streets/
We'll probably go to Dachau on our M|+nchen trip. For obvious reasons.
I went in 1978, but haven't been since, and the missus has never been.
We went to Minneapolis this evening for a couple of nice Czech
("Zizkov") and Slovenian ("Metelkova") Pilsner beers.
Recommended if you're ever in town.
https://www.bauhausbrewlabs.com/
On 28/01/2026 11:18, Champ wrote:
On Wed, 28 Jan 2026 08:58:51 +0000, AndrewR
<andrew.rockface.spam@gmail.com> wrote:
<of AI>
I also invite them to consider if they really think their job would
be better if it was reduced to explaining, repeatedly, what they
wanted the work-experience kid to do, and then checking it was done
right.
Exactly. If the dev job becomes supervising and correcting the
machines, that, imo, is a serious downgrade
As seen elsehere:
Person 1: Wow! AI can produce art!
Person 2: I don't it to produce 'art'. I want it to do the washing
up, so I can produce art
Also seen elsewhere:
Voltaire stated rCLWork saves a man from three great evils: boredom,
vice, and need.rCY So the big problem is that we must solve all 3, not
just the one.
Bruce <07.013@scorecrow.com> wrote in news:309abc82-0af2-4d29-86b1-976073364d66@scorecrow.com:
On 28/01/2026 11:18, Champ wrote:
On Wed, 28 Jan 2026 08:58:51 +0000, AndrewR
<andrew.rockface.spam@gmail.com> wrote:
<of AI>
I also invite them to consider if they really think their job would
be better if it was reduced to explaining, repeatedly, what they
wanted the work-experience kid to do, and then checking it was done
right.
Exactly. If the dev job becomes supervising and correcting the
machines, that, imo, is a serious downgrade
As seen elsehere:
Person 1: Wow! AI can produce art!
Person 2: I don't it to produce 'art'. I want it to do the washing
up, so I can produce art
Also seen elsewhere:
Voltaire stated rCLWork saves a man from three great evils: boredom,
vice, and need.rCY So the big problem is that we must solve all 3, not
just the one.
[random insertion of this post as my reader has deleted the OP]
an insider's view
https://shumer.dev/something-big-is-happening
I sometimes wonder if my son is vulnerable. Works for a company providing engineering services to the national grid. 24/7 emergency cover as well as maintenance and installations.
He already uses AI tools in his back office role on the logistics side.
an insider's view
https://shumer.dev/something-big-is-happening
On Thu, 12 Feb 2026 12:00:21 -0000 (UTC), wessie
<willnotwork@tesco.net> wrote:
an insider's view
https://shumer.dev/something-big-is-happening
Hmm. An insider massively invested in hyping the hype
https://www.linkedin.com/in/mattshumer/
"Matt is the co-founder and CEO of OthersideAI, an applied AI company building the most advanced autocomplete tools in the world, powered by large-scale AI systems like GPT-3. OthersideAI is the company behind HyperWrite, the leading AI autocomplete Chrome extension for
consumers."
That reads like classic pyramid-selling "don't missed the boat!"
patter to me
Champ <neal@champ.org.uk> wrote in news:i9nuok5chnja1p8juem89gfj0sc0hnsmmr@ >4ax.com:
an insider's view
https://shumer.dev/something-big-is-happening
Hmm. An insider massively invested in hyping the hype
https://www.linkedin.com/in/mattshumer/
That reads like classic pyramid-selling "don't missed the boat!"
patter to me
there are a lot of people expressing the same view as the article has >circulated more
On Fri, 13 Feb 2026 17:42:42 -0000 (UTC), wessie
<willnotwork@tesco.net> wrote:
Champ <neal@champ.org.uk> wrote in
news:i9nuok5chnja1p8juem89gfj0sc0hnsmmr@ 4ax.com:
an insider's view
https://shumer.dev/something-big-is-happening
Hmm. An insider massively invested in hyping the hype
https://www.linkedin.com/in/mattshumer/
That reads like classic pyramid-selling "don't missed the boat!"
patter to me
there are a lot of people expressing the same view as the article has >>circulated more
Yeah. David Emmett (MotoGP journalist) also responded to it on
Bluesky yesterday, saying that proposing AI can replace journalists
just reveals a complete misunderstanding of what journalism is. An AI
can probably summarise a press release, but real journalism is about
human relationships
Champ <neal@champ.org.uk> wrote in news:i9nuok5chnja1p8juem89gfj0sc0hnsmmr@ 4ax.com:
On Thu, 12 Feb 2026 12:00:21 -0000 (UTC), wessie
<willnotwork@tesco.net> wrote:
an insider's view
https://shumer.dev/something-big-is-happening
Hmm. An insider massively invested in hyping the hype
https://www.linkedin.com/in/mattshumer/
"Matt is the co-founder and CEO of OthersideAI, an applied AI company
building the most advanced autocomplete tools in the world, powered by
large-scale AI systems like GPT-3. OthersideAI is the company behind
HyperWrite, the leading AI autocomplete Chrome extension for
consumers."
That reads like classic pyramid-selling "don't missed the boat!"
patter to me
there are a lot of people expressing the same view as the article has circulated more
"In summary, the "Annblack Box" is effectively the newer, slightly
refined version of the "M-Vave Black Box" product line."
This turned out to be a load of old bollocks. The two are identical, the ANNBlackbox was simply a renaming to emphasise its use of ANN (a variant
of NAM, Neural Amp Modelling). This was even more obvious when the
Blackbox that arrived had "ANNBLACKBOX" on it. So AI is not trustworthy.
This turned out to be a load of old bollocks.
On 15/02/2026 15:18, Ben Blaney wrote:
On Feb 14, 2026 at 1:59:28 PM EST, "Mike Fleming" <mike@tauzero.co.uk>
wrote:
<snip AI tribulations>
This turned out to be a load of old bollocks.
In cases where the answer is a load of old bollocks, the root cause is
usually
the prompt. Tightening up the parameters of the question, the scope of
enquiry, the acceptable sources (and verification thereof) - that will
improve
the output.
Indeed.
"It's the way I tell 'em"
If anybody (scorecrow Bruce?) is remotely interested in how Claude
'plans' some (extra) coding after being given a PDF example, a text
lookup file and data extracted by SQL from a database table, then here
is an example. BTW a salutary lesson, make a note of where it writes
plans to disc locally (though find is your friend) in case you
accidentally close the CLI window.
https://www.text2kml.co.uk/Claude%20Chats/inherited-inventing- stonebraker.md
There is another side to this, the acceleration of the destruction of
the planet: >https://www.theguardian.com/technology/2026/feb/13/elon-musk-xai-pollution-naacp
On Sat, 14 Feb 2026 13:54:44 +0000, Mike Fleming <mike@tauzero.co.uk>
wrote:
There is another side to this, the acceleration of the destruction of
the planet:
https://www.theguardian.com/technology/2026/feb/13/elon-musk-xai-pollution-naacp
Quite
So it's unreliable, it will *never* do what's promised fot it, and
it's fucking the planet in new worse ways
On 15/02/2026 16:48, Champ wrote:
On Sat, 14 Feb 2026 13:54:44 +0000, Mike Fleming <mike@tauzero.co.uk>
wrote:
There is another side to this, the acceleration of the destruction of
the planet:
https://www.theguardian.com/technology/2026/feb/13/elon-musk-xai-
pollution-naacp
Quite
So it's unreliable, it will *never* do what's promised fot it, and
it's fucking the planet in new worse ways
Perhaps, "it will *never* do *all* that's promised for it".
Valid second point. I must have consumed at least a bath full of water
this afternoon with my little project. Then I ask myself what I might
have been doing that was equally, or even more destructive, if the
weather wasn't so foul.
On 15/02/2026 18:10, Peter Fisher wrote:
On 15/02/2026 16:48, Champ wrote:
On Sat, 14 Feb 2026 13:54:44 +0000, Mike Fleming <mike@tauzero.co.uk>
wrote:
There is another side to this, the acceleration of the destruction of
the planet:
https://www.theguardian.com/technology/2026/feb/13/elon-musk-xai-
pollution-naacp
Quite
So it's unreliable, it will *never* do what's promised fot it, and
it's fucking the planet in new worse ways
Perhaps, "it will *never* do *all* that's promised for it".
Valid second point. I must have consumed at least a bath full of water
this afternoon with my little project. Then I ask myself what I might
have been doing that was equally, or even more destructive, if the
weather wasn't so foul.
Well, you could have had a nice relaxing soak in the bath?
On 16/02/2026 08:59, YTC1 wrote:
On 15/02/2026 18:10, Peter Fisher wrote:
On 15/02/2026 16:48, Champ wrote:
On Sat, 14 Feb 2026 13:54:44 +0000, Mike Fleming <mike@tauzero.co.uk>
wrote:
There is another side to this, the acceleration of the destruction of >>>>> the planet:
https://www.theguardian.com/technology/2026/feb/13/elon-musk-xai-
pollution-naacp
Quite
So it's unreliable, it will *never* do what's promised fot it, and
it's fucking the planet in new worse ways
Perhaps, "it will *never* do *all* that's promised for it".
Valid second point. I must have consumed at least a bath full of
water this afternoon with my little project. Then I ask myself what I
might have been doing that was equally, or even more destructive, if
the weather wasn't so foul.
Well, you could have had a nice relaxing soak in the bath?
Not without going to the swimming pool.
Walk in shower only here (with a fold down seat and a 'rain head' for something approximating a soak).
On 15/02/2026 15:58, Peter Fisher wrote:
On 15/02/2026 15:18, Ben Blaney wrote:
On Feb 14, 2026 at 1:59:28 PM EST, "Mike Fleming"
<mike@tauzero.co.uk> wrote:
<snip AI tribulations>
This turned out to be a load of old bollocks.
In cases where the answer is a load of old bollocks, the root cause
is usually
the prompt. Tightening up the parameters of the question, the scope of
enquiry, the acceptable sources (and verification thereof) - that
will improve
the output.
Indeed.
"It's the way I tell 'em"
If anybody (scorecrow Bruce?) is remotely interested in how Claude
'plans' some (extra) coding after being given a PDF example, a text
lookup file and data extracted by SQL from a database table, then here
is an example. BTW a salutary lesson, make a note of where it writes
plans to disc locally (though find is your friend) in case you
accidentally close the CLI window.
https://www.text2kml.co.uk/Claude%20Chats/inherited-inventing-
stonebraker.md
Yes, I know that there is a fault in the cunning plan. Top Ten
qualifiers and hence runners may well be on a 3-wheeler. IIRC it
actually sought clarification on that, but I forgot to answer - like Ben said up there.
"Tightening up the parameters of the question, the scope of enquiry, the acceptable sources (and verification thereof) - that will improve the output."
Anyway, using Claude has been described as like talking to a junior dev.
On 15/02/2026 16:13, Peter Fisher wrote:<snip>
On 15/02/2026 15:58, Peter Fisher wrote:
On 15/02/2026 15:18, Ben Blaney wrote:
On Feb 14, 2026 at 1:59:28 PM EST, "Mike Fleming"
<mike@tauzero.co.uk> wrote:
<snip AI tribulations>
This turned out to be a load of old bollocks.
In cases where the answer is a load of old bollocks, the root cause
is usually
the prompt. Tightening up the parameters of the question, the scope of >>>> enquiry, the acceptable sources (and verification thereof) - that
will improve
the output.
"Tightening up the parameters of the question, the scope of enquiry,
the acceptable sources (and verification thereof) - that will improve
the output."
The issue I have with this is that it only has validity if you *know*
that the answer is "a load of old bollocks". Further, it only has
validity if you know to (ie ...have the 'common sense' to...) check the output is an accurate representation of the paradigm[1].
Anyway, using Claude has been described as like talking to a junior dev.
I admit, I don't even meet that criterion :-)
Otherwise, I'm with what Champ said somewhere up there^ that we're just finding new ways to f*uck up the planet.-a And probably faster than ever.
On 14/02/2026 18:59, Mike Fleming wrote:
"In summary, the "Annblack Box" is effectively the newer, slightly
refined version of the "M-Vave Black Box" product line."
This turned out to be a load of old bollocks. The two are identical,
the ANNBlackbox was simply a renaming to emphasise its use of ANN (a
variant of NAM, Neural Amp Modelling). This was even more obvious when
the Blackbox that arrived had "ANNBLACKBOX" on it. So AI is not
trustworthy.
Which LLM?
"The M-VAVE Black Box and M-VAVE ANN Black Box are essentially the same multi-effects processor for guitar and bass, with the "ANN" in the
latter name emphasizing its Audio Neural Network modelling technology.
The product was likely renamed in some regions to avoid legal issues
with other products named "Black Box".
The choice between the M-VAVE Black Box and M-VAVE ANN Black Box is a
matter of finding the best current price, as the features and
performance are identical across the product name variations."
On 14/02/2026 18:59, Mike Fleming wrote:
I'd asked if anyone on a bass forum knew the answer, and someone
resorted to AI,
This, I find so patronising. Like you couldn't ask AI for yourself.
And no one who ever does that says "and I've verified the response".
If anybody (scorecrow Bruce?) is remotely interested in how Claude
'plans' some (extra) coding after being given a PDF example, a text
lookup file and data extracted by SQL from a database table, then here
is an example. BTW a salutary lesson, make a note of where it writes
plans to disc locally (though find is your friend) in case you
accidentally close the CLI window.
https://www.text2kml.co.uk/Claude%20Chats/inherited-inventing-stonebraker.md
On 15/02/2026 15:58, Peter Fisher wrote:
If anybody (scorecrow Bruce?) is remotely interested in how Claude
'plans' some (extra) coding after being given a PDF example, a text
lookup file and data extracted by SQL from a database table, then here
is an example. BTW a salutary lesson, make a note of where it writes
plans to disc locally (though find is your friend) in case you
accidentally close the CLI window.
https://www.text2kml.co.uk/Claude%20Chats/inherited-inventing-
stonebraker.md
I was nearly about to try a Claude subscription on a test project to see
for myself how effective it is, when I read this:
<https://www.theregister.com/2026/02/16/anthropic_claude_ai_edits/>
You'd think they would be focusing on retaining existing customers and acquiring new ones? But no, apparently they have now reached the stage
of "we'll dick our users around just because we can".
(Perhaps their devs were all recruited from the Gnome desktop team?)
On 15/02/2026 15:58, Peter Fisher wrote:
If anybody (scorecrow Bruce?) is remotely interested in how Claude
'plans' some (extra) coding after being given a PDF example, a text
lookup file and data extracted by SQL from a database table, then here
is an example. BTW a salutary lesson, make a note of where it writes
plans to disc locally (though find is your friend) in case you
accidentally close the CLI window.
https://www.text2kml.co.uk/Claude%20Chats/inherited-inventing-
stonebraker.md
I was nearly about to try a Claude subscription on a test project to see
for myself how effective it is, when I read this:
<https://www.theregister.com/2026/02/16/anthropic_claude_ai_edits/>
You'd think they would be focusing on retaining existing customers and acquiring new ones? But no, apparently they have now reached the stage
of "we'll dick our users around just because we can".
(Perhaps their devs were all recruited from the Gnome desktop team?)
| Sysop: | Amessyroom |
|---|---|
| Location: | Fayetteville, NC |
| Users: | 59 |
| Nodes: | 6 (0 / 6) |
| Uptime: | 24:33:14 |
| Calls: | 810 |
| Files: | 1,287 |
| Messages: | 195,978 |