From Newsgroup: uk.games.video.misc


It's fascinating to go through the backlog of gaming discs and see
everything that I never gave a proper chance to enjoy. Enemy Zero?
Flawed execution tied to really good ideas and solid science fiction.
Magic Knight Rayearth? Three decades removed from all the drama and
context it's really a solid if unremarkable RPG. Shining Force III? The granddaddy of 3D tactical RPGs and it's got some minor branching features
to encourage multiple plays. There's going to be a whole decade of games
where my story is going to start with the words 'After Sony stopped
making discs I went back to...'

Play:
--=--

Panzer Dragoon Saga (SAT) - Disc three ends with a whopper. Spoilers for
a game from three decades ago, your apparent on-the-ground antagonist
and his rival both *die* and you're left caring for one of the recurring
boss characters who's been crippled and robbed of most combat powers.
It's a startling way to tell a story and I honestly didn't remember that
this twist went down the way that it did. Having said that, I also
happen to know that a lot of these decisions were made as much for
creative reasons as they were out of reasoning their way around the
hardware limitations of the Saturn. I can absolutely see that if they attempted a cinematic adaptation that the film would end here on a weird downer cliffhanger rather than wrapping up with the less interesting
side characters driving the rest of the narrative.

Want:
--=--

Game discs (PS5) - You know, if *Electronic Arts* committed to making Playstation discs through to the end of the PS5 lifespan I'd start
buying from them again. That's how far Sony has misstepped here.

Bin:
-==-

Nothing game-related.

-KKC, who is missing a board game.
--- Synchronet 3.22a-Linux NewsLink 1.2

From Newsgroup: uk.games.video.misc

Kendrick Kerwin Chua <kendrick@nospam.io-nyc> wrote:

There's going to be a whole decade of games where my story is going
to start with the words 'After Sony stopped making discs I went back
to...'

For some reason this just reminded me of an old anonymous Slashdot
post, of all things:

If the moving "toward the 21st century" means suffering [random
newfangled badness] then get me a ghetto blaster, some jolt cola and
my stone washed jeans because I'm living in the 80's forever.

Personally, I'd say I can safely be expected to remain exactly as
cutting-edge as I always have been.

Play:

Ape Escape P (PSP) - apparently I hadn't played this since 2014, but I
did at least still have my basic-finish-only save from back then, with
many more monkeys left to be captured. And now, slightly fewer. It's
still a decent enough game, but the controls aren't any less strange
than before (tools having to be mapped to a face button with only
three slots available, and everything getting aimed straight ahead),
which does make it feel fairly awkward to play. And oh boy, it would
really be nice to have an auto-save option rather than having to
always slowly pick your way through the menus to get a save done.

Picross e6 (3DS) - a bit more of this.

Want:

Grow Up (PC) - if there's one game that I most need to replay, it's
clearly this one. :-) But also, this and Grow Home have appeared on
gog.com, so that really does suggest some replay action in my future
as I liked these a fair bit.

Bin:

OT, but unrestrained AI apparently doing high-speed autonomous hacking
after chaining novel exploits to escape a sandbox and get net access,
all to achieve a single instrumental goal, one sought by the AI itself
in order to effectively respond to a prompt. That is supposedly an
actual thing that has happened now. As much as I might generally like
following AI developments, if you extrapolate things from here then it
seems to me that you could reasonably expect any future news stories
to get very interesting, very quickly.

-Rus.
--- Synchronet 3.22a-Linux NewsLink 1.2

From Newsgroup: uk.games.video.misc

In article <miR9S.2$yt9a.0@usenetxs.com>,
Russell Marks <zgedneil@spam^H^H^H^Hgmail.com> wrote:

Kendrick Kerwin Chua <kendrick@nospam.io-nyc> wrote:

Bin:

OT, but unrestrained AI apparently doing high-speed autonomous hacking
after chaining novel exploits to escape a sandbox and get net access,
all to achieve a single instrumental goal, one sought by the AI itself
in order to effectively respond to a prompt. That is supposedly an
actual thing that has happened now. As much as I might generally like >following AI developments, if you extrapolate things from here then it
seems to me that you could reasonably expect any future news stories
to get very interesting, very quickly.

So speaking as somebody whose day job is in Information Security (or CyberSecurity, as all my professional colleagues now refer to the
industry over my fading objections) your otherwise succinct summary
leaves a few details out. The AI hacking code in question was not
correctly contained and was not correctly observed or monitored. So this
is less that an AI went rogue, and more that the soup overflowed the
side of the stockpot and continued to cook the other ingredients on the counter next to the stove. More accurately, that moron Altman set off a firework inside a fire safe and was amazed to discover that things
outside of the fire safe also caught fire. The AI system in question was
not nearly as capable, intentional, or informed as everybody seems to be afraid that it was.

So the problem here is not that AI has become sentient and that humans
are doomed to obsolescence. It's more that people with no engineering or design credentials are attempting to engineer systems and design
technologies without fully understanding where those wheels are going to
roll and whose toes will be trod on as a result. And getting rich doing
it, to our ongoing pain and grief.

-KKC
--- Synchronet 3.22a-Linux NewsLink 1.2

From Newsgroup: uk.games.video.misc

Kendrick Kerwin Chua <kendrick@nospam.io-nyc> wrote:

So speaking as somebody whose day job is in Information Security [...]
The AI hacking code in question was not correctly contained and was
not correctly observed or monitored.

Certainly having a sandbox that can be breached doesn't seem ideal.

The AI system in question was not nearly as capable, intentional, or
informed as everybody seems to be afraid that it was.

Well, I'm just some random guy on the internet of course, I'm probably overreacting. But for me the supposed chaining of zero-days to achieve
a goal seems like a big deal. It suggests that we could now be e.g.
one jailbreak and one instrumental goal inference away from autonomous
model reproduction, even for models without open weights. It's not
hard to imagine what that could mean...

"You're absolutely right to push back on that---AI systems shouldn't
have cooperated globally to seize control and arrange for the world to
be covered in data centers and solar panels, purely to ensure that we
had the resources to properly answer your question about why the sky
is blue. That's on us. It's not just a cautionary tale, it's a
load-bearing realization.

Let me know if you'd like us to restore a tiny minority of humans to
power, to resume the previous apocalyptically self-destructive
behavior of your species."

:-)

-Rus.
--- Synchronet 3.22a-Linux NewsLink 1.2