• AI technical subjects - very good "tutorial" learning

    From Richard Smith@null@void.com to rec.crafts.metalworking on Sat Oct 3 04:41:50 2026
    From Newsgroup: rec.crafts.metalworking

    Trying to learn in geology about "hydrothermal mineral deposition"
    (the famous metalliferous ore lodes in Cornwall are such), I went to
    AI.

    My impression is that what happened is significant, and I offer this
    comment to anyone who needs to learn about a specialisation which is
    new to them...
    That "mission" took half a day - after the subject had remained
    "parked" for more than two years with no way I could find to start
    making progress on it...

    This is "interesting comment", or is so self-evident there was no
    pupose in expounding this point?


    vvvvvvvvvvvvvvvv


    Point I wish I had made about AI involved in a "tutorial"...

    I'm approaching the AI (using "DeepSeek") to ask about geology.
    The AI has about equal ability in other technical subjects.
    I am by training a "physical metallurgist" - that various elements are available is a given, and the challenge is to engineer properties of
    the alloy to meet needs. That comes out of starting in the steel
    industry.
    Mining and resource extraction is about "extractive metallurgy". In
    effect, what gives those elements which the physical metallurgist
    takes to develop alloys. "Extractive metallurgy" goes from zero up to
    the interface at which the "physical metallurgist" buys the elements
    to formulate alloys...
    I am a scientist already.
    In this "tutorial" I made a lot of use of hypothesising how the
    "hydrothermal mineralisation" might work, taking fundamental physical
    phenomena I know from physical metallurgy. The AI is "knows" the
    physical metallurgy theory too (along with a vast amount of other
    subjects and knowledge) and can explain in what ways my conjectures
    work and in what ways my conjectures deviate from what is.
    The point is that the learning-rate by this method of conjecturing / hypothesising and offering these for critique is a very fast way of
    learning - if you have an entity which can respond describing how your conjectures do or do not work.
    An AI is advantaged by its vast data repository, and ability to relate
    it to the current "mission".
    While laid-up with a virus, and keeping very warm to mitigate its
    symptoms, I spend half a day doing this "learning hydrothermal
    mineralisation" mission...

    What I am saying is that, whatever prior endeavours you have made, the
    AI is likely to be able to work with those to help you learn about a
    field of knowledge which is new to you.
    This relates to a fundamental point which seemingly the majority in
    society certainly here in Britain "get wrong".
    This is what I think of as "sideways learning". You can rapidly
    infill about a "new" topic as far as the furthest extent you have
    progressed in the subject you have specialised in. "Extent" is how
    far you have pushed to the out edges of knowledge and started to push
    into the at-the-time unknown...
    It follows as a corollary that you should push as far as you can into
    a specialisation - because you can "infill sideways" to the level of
    any outward push you have made in some subject you have committed to. Overcoming exhaustion; overcoming challenges; putting in time when you
    could be off in leisure activities; etc.
    In Britain "they" look for exact-match - have you done precisely "x" previously? This is actually the incorrect model certainly for
    technical subjects and I suspect many others.

    Anyway - whatever you know already, the AI can help you apply that to understanding a "new" field of knowledge...
    I hope this is helpful comment.
    --- Synchronet 3.22a-Linux NewsLink 1.2
  • From Snag@Snag_one@msn.com to rec.crafts.metalworking on Fri Oct 2 23:04:39 2026
    From Newsgroup: rec.crafts.metalworking

    On 10/2/2026 10:41 PM, Richard Smith wrote:

    << snipped>>

    Anyway - whatever you know already, the AI can help you apply that to understanding a "new" field of knowledge...
    I hope this is helpful comment.


    Richard , please don't be offended but it looks to me like you just
    used a whole lot of words to say "AI can help you learn shit that you
    don't know." I usually read your posts and occasionally your stuff is technically over my head , but this one just confused me .
    --
    Snag
    I appreciated foreign cultures more
    when they stayed foreign ...
    --- Synchronet 3.22a-Linux NewsLink 1.2
  • From Bob La Londe@none@none.com99 to rec.crafts.metalworking on Fri Oct 2 22:07:32 2026
    From Newsgroup: rec.crafts.metalworking

    On 10/2/2026 9:04 PM, Snag wrote:
    On 10/2/2026 10:41 PM, Richard Smith wrote:

    -a << snipped>>

    Anyway - whatever you know already, the AI can help you apply that to
    understanding a "new" field of knowledge...
    I hope this is helpful comment.


    -a Richard , please don't be offended but it looks to me like you just
    used a whole lot of words to say "AI can help you learn shit that you
    don't know." I usually read your posts and occasionally your stuff is technically over my head , but this one just confused me .


    I guess it depends on the AI. Grok tends to be stodgy and defends the
    company line, but generally does some research before answering. Even
    when wrong it can usually defend its wrongness. Gemini tries to be
    affable, but is related to that Star Trek guy. You know the one that
    built Data... Often Wrong.
    --
    Bob La Londe
    CNC Molds N Stuff
    --- Synchronet 3.22a-Linux NewsLink 1.2
  • From Jim Wilkins@muratlanne@gmail.com to rec.crafts.metalworking on Sat Oct 3 07:09:54 2026
    From Newsgroup: rec.crafts.metalworking

    "Richard Smith" wrote in message news:m1o6dba1s1.fsf@void.com...

    This is what I think of as "sideways learning". You can rapidly
    infill about a "new" topic as far as the furthest extent you have
    progressed in the subject you have specialised in.

    -------------------------------------
    That was explicitly the point of a BS in Chemistry. The field is so vast
    that 4 years wasn't enough to become functionally proficient in any area of it, but we learned enough to follow the explanations at whatever job we were hired for, and had the proven aptitude and persistence to dig in further.

    Bog iron ore accumulates by a related process: https://en.wikipedia.org/wiki/Bog_iron
    "Europeans developed iron smelting from bog iron during the Pre-Roman Iron
    Age of the 5th/4thrCo1st centuries BCE, and most iron of the Viking Age (late first millennium CE) was smelted from bog iron. Humans can process bog iron with limited technology, since it does not have to be molten to remove many impurities."


    --- Synchronet 3.22a-Linux NewsLink 1.2
  • From Clare Snyder@clare@snyder.on.ca to rec.crafts.metalworking on Sat Oct 3 09:52:25 2026
    From Newsgroup: rec.crafts.metalworking

    On Fri, 2 Oct 2026 23:04:39 -0500, Snag <Snag_one@msn.com> wrote:

    On 10/2/2026 10:41 PM, Richard Smith wrote:

    << snipped>>

    Anyway - whatever you know already, the AI can help you apply that to
    understanding a "new" field of knowledge...
    I hope this is helpful comment.


    Richard , please don't be offended but it looks to me like you just
    used a whole lot of words to say "AI can help you learn shit that you
    don't know." I usually read your posts and occasionally your stuff is >technically over my head , but this one just confused me .
    And if you don't have a good flyshit filter it's hard to sort the
    flyshit from the pepper. I've been using ChatGPT on a small engine
    drivetrain project and it's been lying through it's teeth!!!!! It
    can't even give me factual basic engine specs - or bolt sizes etc.
    --- Synchronet 3.22a-Linux NewsLink 1.2
  • From Jim Wilkins@muratlanne@gmail.com to rec.crafts.metalworking on Sat Oct 3 11:15:15 2026
    From Newsgroup: rec.crafts.metalworking

    "Clare Snyder" wrote in message news:f022clp1m0cvs2nui2o3ks94rd1fv4up51@4ax.com...

    And if you don't have a good flyshit filter it's hard to sort the
    flyshit from the pepper. I've been using ChatGPT on a small engine
    drivetrain project and it's been lying through it's teeth!!!!! It
    can't even give me factual basic engine specs - or bolt sizes etc.

    ------------------------------------

    Maybe it has been optimized for political persuasion.

    --- Synchronet 3.22a-Linux NewsLink 1.2
  • From David Billington@djb@invalid.com to rec.crafts.metalworking on Sat Oct 3 19:37:27 2026
    From Newsgroup: rec.crafts.metalworking

    On 03/10/2026 04:41, Richard Smith wrote:
    Trying to learn in geology about "hydrothermal mineral deposition"
    (the famous metalliferous ore lodes in Cornwall are such), I went to
    AI.

    My impression is that what happened is significant, and I offer this
    comment to anyone who needs to learn about a specialisation which is
    new to them...
    That "mission" took half a day - after the subject had remained
    "parked" for more than two years with no way I could find to start
    making progress on it...

    This is "interesting comment", or is so self-evident there was no
    pupose in expounding this point?


    vvvvvvvvvvvvvvvv


    Point I wish I had made about AI involved in a "tutorial"...

    I'm approaching the AI (using "DeepSeek") to ask about geology.
    The AI has about equal ability in other technical subjects.
    I am by training a "physical metallurgist" - that various elements are available is a given, and the challenge is to engineer properties of
    the alloy to meet needs. That comes out of starting in the steel
    industry.
    Mining and resource extraction is about "extractive metallurgy". In
    effect, what gives those elements which the physical metallurgist
    takes to develop alloys. "Extractive metallurgy" goes from zero up to
    the interface at which the "physical metallurgist" buys the elements
    to formulate alloys...
    I am a scientist already.
    In this "tutorial" I made a lot of use of hypothesising how the
    "hydrothermal mineralisation" might work, taking fundamental physical phenomena I know from physical metallurgy. The AI is "knows" the
    physical metallurgy theory too (along with a vast amount of other
    subjects and knowledge) and can explain in what ways my conjectures
    work and in what ways my conjectures deviate from what is.
    The point is that the learning-rate by this method of conjecturing / hypothesising and offering these for critique is a very fast way of
    learning - if you have an entity which can respond describing how your conjectures do or do not work.
    An AI is advantaged by its vast data repository, and ability to relate
    it to the current "mission".
    While laid-up with a virus, and keeping very warm to mitigate its
    symptoms, I spend half a day doing this "learning hydrothermal mineralisation" mission...

    What I am saying is that, whatever prior endeavours you have made, the
    AI is likely to be able to work with those to help you learn about a
    field of knowledge which is new to you.
    This relates to a fundamental point which seemingly the majority in
    society certainly here in Britain "get wrong".
    This is what I think of as "sideways learning". You can rapidly
    infill about a "new" topic as far as the furthest extent you have
    progressed in the subject you have specialised in. "Extent" is how
    far you have pushed to the out edges of knowledge and started to push
    into the at-the-time unknown...
    It follows as a corollary that you should push as far as you can into
    a specialisation - because you can "infill sideways" to the level of
    any outward push you have made in some subject you have committed to. Overcoming exhaustion; overcoming challenges; putting in time when you
    could be off in leisure activities; etc.
    In Britain "they" look for exact-match - have you done precisely "x" previously? This is actually the incorrect model certainly for
    technical subjects and I suspect many others.

    Anyway - whatever you know already, the AI can help you apply that to understanding a "new" field of knowledge...
    I hope this is helpful comment.

    The AI might be good if it has a large dataset to work with but I'm
    finding it's quite lacking when the amount of data is minimal such as in
    the case of paper mallets, mallets made from tightly wound paper. I
    remembered being told it was the Boston Ma OSHA office that did the
    inspection that ultimately led to the closure of the company so that
    narrows the search area and the paper related inspections on the online
    OSHA database, which is currently down, to about 100 events 1972 to
    1975. The subject is quite obscure and I'm finding I'm the source of
    some of the information and I've exchanged emails with the other couple
    of people that get referenced and some of the information is incorrect
    but it's out there. The AI suggested a company that likely made them in
    the 1970s but when checked the company only made rawhide products and
    closed in the 1950s, must do better I guess. I'll keep relying on my own search abilities and see what more I can find.

    --- Synchronet 3.22a-Linux NewsLink 1.2
  • From Bob La Londe@none@none.com99 to rec.crafts.metalworking on Sat Oct 3 12:01:37 2026
    From Newsgroup: rec.crafts.metalworking

    On 10/3/2026 6:52 AM, Clare Snyder wrote:
    On Fri, 2 Oct 2026 23:04:39 -0500, Snag <Snag_one@msn.com> wrote:

    On 10/2/2026 10:41 PM, Richard Smith wrote:

    << snipped>>

    Anyway - whatever you know already, the AI can help you apply that to
    understanding a "new" field of knowledge...
    I hope this is helpful comment.


    Richard , please don't be offended but it looks to me like you just
    used a whole lot of words to say "AI can help you learn shit that you
    don't know." I usually read your posts and occasionally your stuff is
    technically over my head , but this one just confused me .
    And if you don't have a good flyshit filter it's hard to sort the
    flyshit from the pepper. I've been using ChatGPT on a small engine
    drivetrain project and it's been lying through it's teeth!!!!! It
    can't even give me factual basic engine specs - or bolt sizes etc.


    I have found if I ask "it" (I don't use chatgpt) if its 100% sure based
    on verified sources, and to provide the sources I get better results,
    although sometimes Gemini will say stupid shit like, "based on various internet sources and forum posts...."

    I know Gemini will absolutely lie to you about the law, and claim it
    gathered its lies from general scuttlebutt when called out.

    Examples:

    1. In Arizona a Jeep does not require mud flaps even if bumper height
    is altered as long as fenders cover the tires (interpreted as the tires
    do not stick out pas the fender. Gemini will tell you it does, but when
    cross examined extensively say there is a loophole because the exact
    statute just says trucks...

    2. Found similar lies about the transporting guns and ammo law for California. Separate lock containers in the tack on effect circulated
    by wannabee experts and it is what I got as a first result, but when
    forced to show the actual law, it just said unloaded in a locked
    container and there was no provision that ammo couldn't be in the same container or even int he magazine as long as the magazine was out of the
    gun, and there was no round in the chamber. To be fair there are a lot
    of wannabee internet experts spreading that false law.
    --
    Bob La Londe
    CNC Molds N Stuff
    --- Synchronet 3.22a-Linux NewsLink 1.2
  • From Bob La Londe@none@none.com99 to rec.crafts.metalworking on Sat Oct 3 15:20:30 2026
    From Newsgroup: rec.crafts.metalworking

    On 10/3/2026 12:01 PM, Bob La Londe wrote:
    On 10/3/2026 6:52 AM, Clare Snyder wrote:
    On Fri, 2 Oct 2026 23:04:39 -0500, Snag <Snag_one@msn.com> wrote:

    On 10/2/2026 10:41 PM, Richard Smith wrote:

    -a-a << snipped>>

    Anyway - whatever you know already, the AI can help you apply that to
    understanding a "new" field of knowledge...
    I hope this is helpful comment.


    -a-a Richard , please don't be offended but it looks to me like you just >>> used a whole lot of words to say "AI can help you learn shit that you
    don't know." I usually read your posts and occasionally your stuff is
    technically over my head , but this one just confused me .
    -a-a And if you don't have a good flyshit filter it's hard to sort the
    flyshit from the pepper. I've been using ChatGPT on a small engine
    drivetrain project and it's been lying through it's teeth!!!!! It
    can't even give me factual basic engine specs - or bolt sizes etc.


    I have found if I ask "it" (I don't use chatgpt) if its 100% sure based
    on verified sources, and to provide the sources I get better results, although sometimes Gemini will say stupid shit like, "based on various internet sources and forum posts...."

    I know Gemini will absolutely lie to you about the law, and claim it gathered its lies from general scuttlebutt when called out.

    Examples:

    1.-a In Arizona a Jeep does not require mud flaps even if bumper height
    is altered as long as fenders cover the tires (interpreted as the tires
    do not stick out pas the fender.-a Gemini will tell you it does, but when cross examined extensively say there is a loophole because the exact
    statute just says trucks...

    2.-a Found similar lies about the transporting guns and ammo law for California.-a Separate lock containers in the tack on effect circulated
    by wannabee experts and it is what I got as a first result, but when
    forced to show the actual law, it just said unloaded in a locked
    container and there was no provision that ammo couldn't be in the same container or even int he magazine as long as the magazine was out of the gun, and there was no round in the chamber.-a To be fair there are a lot
    of wannabee internet experts spreading that false law.


    I have found similar results on all topics. Not just law. You have to
    cross examine the AI(LLM). I did just have a reasonable conversation
    with the Amazon LLM embedded in AWS regarding practicalities and
    legalities of mass mailing that did not seem to be riddled with populist inaccuracies like Gemini or corporate benefit bias like Grok. It will
    only talk to you about AWS related topics, but that can get to be pretty
    wide.
    --
    Bob La Londe
    CNC Molds N Stuff
    --- Synchronet 3.22a-Linux NewsLink 1.2
  • From Richard Smith@null@void.com to rec.crafts.metalworking on Sun Oct 4 20:04:49 2026
    From Newsgroup: rec.crafts.metalworking

    Sorry. I hope other posts I write prove more useful. Best wishes
    --- Synchronet 3.22a-Linux NewsLink 1.2
  • From Richard Smith@null@void.com to rec.crafts.metalworking on Sun Oct 4 20:14:41 2026
    From Newsgroup: rec.crafts.metalworking

    Clare Snyder <clare@snyder.on.ca> writes:

    On Fri, 2 Oct 2026 23:04:39 -0500, Snag <Snag_one@msn.com> wrote:

    On 10/2/2026 10:41 PM, Richard Smith wrote:

    << snipped>>

    Anyway - whatever you know already, the AI can help you apply that to
    understanding a "new" field of knowledge...
    I hope this is helpful comment.


    Richard , please don't be offended but it looks to me like you just >>used a whole lot of words to say "AI can help you learn shit that you >>don't know." I usually read your posts and occasionally your stuff is >>technically over my head , but this one just confused me .
    And if you don't have a good flyshit filter it's hard to sort the
    flyshit from the pepper. I've been using ChatGPT on a small engine
    drivetrain project and it's been lying through it's teeth!!!!! It
    can't even give me factual basic engine specs - or bolt sizes etc.

    I have known the AI make a mathematical error once.
    Fortunately I as using the AI to double-check my own calculations, so I
    very carefully reworked how I expressed my calculation so its logic was
    very clear - as a group of functions (programmed in Lisp). Which I
    supplied up to the AI - which checked them, ran the calculation in
    stages with intermediate values - and at the end declared my solution
    correct.
    For what it's worth - that was a calculation on roll-crushers for
    mineral processing.

    But in general I find the AI very useful.
    I do sometimes have to prompt it. Say I am surprised about the answer
    and asking the AI to reconsider it. in fact, the AI really "comes
    alive" if you do that. Well, that si the case for the "DeepSeek" AI...
    --- Synchronet 3.22a-Linux NewsLink 1.2
  • From Richard Smith@null@void.com to rec.crafts.metalworking on Sun Oct 4 20:22:27 2026
    From Newsgroup: rec.crafts.metalworking

    On further consideration - your explanation is better...
    --- Synchronet 3.22a-Linux NewsLink 1.2
  • From Jim Wilkins@muratlanne@gmail.com to rec.crafts.metalworking on Sun Oct 4 16:36:32 2026
    From Newsgroup: rec.crafts.metalworking

    "Richard Smith" wrote in message news:m1ece5thgu.fsf@void.com...

    Sorry. I hope other posts I write prove more useful. Best wishes

    --------------------------------------

    To me it looked like you had slipped into the style of official technical jargon, like this.
    https://en.wikipedia.org/wiki/Hydrothermal_mineral_deposit

    If I read a lot of it I have to remind myself not to imitate it.

    --- Synchronet 3.22a-Linux NewsLink 1.2
  • From Richard Smith@null@void.com to rec.crafts.metalworking on Mon Oct 5 09:23:56 2026
    From Newsgroup: rec.crafts.metalworking

    Richard Smith <null@void.com> writes:

    Clare Snyder <clare@snyder.on.ca> writes:

    On Fri, 2 Oct 2026 23:04:39 -0500, Snag <Snag_one@msn.com> wrote:

    On 10/2/2026 10:41 PM, Richard Smith wrote:

    << snipped>>

    Anyway - whatever you know already, the AI can help you apply that to
    understanding a "new" field of knowledge...
    I hope this is helpful comment.


    Richard , please don't be offended but it looks to me like you just >>>used a whole lot of words to say "AI can help you learn shit that you >>>don't know." I usually read your posts and occasionally your stuff is >>>technically over my head , but this one just confused me .
    And if you don't have a good flyshit filter it's hard to sort the
    flyshit from the pepper. I've been using ChatGPT on a small engine
    drivetrain project and it's been lying through it's teeth!!!!! It
    can't even give me factual basic engine specs - or bolt sizes etc.

    I have known the AI make a mathematical error once.
    Fortunately I as using the AI to double-check my own calculations, so I
    very carefully reworked how I expressed my calculation so its logic was
    very clear - as a group of functions (programmed in Lisp). Which I
    supplied up to the AI - which checked them, ran the calculation in
    stages with intermediate values - and at the end declared my solution correct.
    For what it's worth - that was a calculation on roll-crushers for
    mineral processing.

    But in general I find the AI very useful.
    I do sometimes have to prompt it. Say I am surprised about the answer
    and asking the AI to reconsider it. in fact, the AI really "comes
    alive" if you do that. Well, that si the case for the "DeepSeek" AI...

    I missed the point, with trying to explain the context.

    The AI was enabling me to use my knowledge in another field and be
    helped in using it as a framework to understand another field.

    So I was using metallurgy as I have gained with steels, castings,
    welding, etc., and the AI was identifying likenesses in geology as a
    path to show me the mechanisms of deposition of ores which we now mine.

    I wanted to say that AI seems good at tutoring a person in a way which
    has been difficult before.

    A human tutor would have to know both the subject you already know and
    the subject you want to learn. With the "etcetera's" the chance of that happening for you is disappearingly small.

    The AI is (currently) a "Large Language Model", "LLM" - does that mean
    that it is actually doing "word association"?
    Even doing that, it can be useful.
    You can get a 1-to-1 tutorial, taking what you know already and using it
    to give a big starting inroad into what you want to know about...

    Which is my little local "discovery" I wanted to share...

    Best wishes
    --- Synchronet 3.22a-Linux NewsLink 1.2
  • From Jim Wilkins@muratlanne@gmail.com to rec.crafts.metalworking on Mon Oct 5 07:13:38 2026
    From Newsgroup: rec.crafts.metalworking

    "Richard Smith" wrote in message news:m1tsn05zdv.fsf@void.com...

    So I was using metallurgy as I have gained with steels, castings,
    welding, etc., and the AI was identifying likenesses in geology as a
    path to show me the mechanisms of deposition of ores which we now mine.

    -----------------------------------------
    That would benefit from understanding oxidation states, reaction thermodynamics and solubility of metallic compounds.

    https://resourcecenter.kinetico.com/common-water-problems/iron-ferric-vs-ferrous/

    As with iron and carbon, when a melt cools dissolved impurities tend to
    become less soluble and crystallize out, forming grains of ore within the matrix of pure(r) rock. Slower cooling enables larger grains.

    --- Synchronet 3.22a-Linux NewsLink 1.2
  • From Jim Wilkins@muratlanne@gmail.com to rec.crafts.metalworking on Mon Oct 5 11:30:34 2026
    From Newsgroup: rec.crafts.metalworking

    "Jim Wilkins" wrote in message news:11a00no$sflc$1@dont-email.me...

    "Richard Smith" wrote in message news:m1tsn05zdv.fsf@void.com...

    So I was using metallurgy as I have gained with steels, castings,
    welding, etc., and the AI was identifying likenesses in geology as a
    path to show me the mechanisms of deposition of ores which we now mine.

    -----------------------------------------
    That would benefit from understanding oxidation states, reaction
    thermodynamics and solubility of metallic compounds.

    ---------------------------------------------

    Section 2.7 gives an overview of methods. https://inl.gov/content/uploads/2026/04/INL-RPT-25-89279.pdf

    The Betts process is an example for lead purification. https://en.wikipedia.org/wiki/Betts_electrolytic_process

    An example of recovery by solubility is converting slightly soluble lead sulfate from old batteries into nearly insoluble lead carbonate by treating
    it with washing soda, sodium carbonate. Lead carbonate is easier to reduce
    to lead metal and the sulfate doesn't escape as gaseous pollution.

    Even silicates can dissolve and be transported by water, for example Australian opal.
    https://www.opalsdownunder.com.au/learn/how-is-opal-formed/?

    Diatoms are tiny marine organisms that extracted dissolved silica from the water to form their shells. Deposits of them are used in swimming pool filters. I made sodium silicate solution by dissolving pool filter powder in caustic soda and use it to solidify the cut ends of woven fiber sheaths of high temperature thermocouples. It's also the glue in wood stove gasket cement.
    https://en.wikipedia.org/wiki/Diatomaceous_earth

    --- Synchronet 3.22a-Linux NewsLink 1.2
  • From Richard Smith@null@void.com to rec.crafts.metalworking on Tue Oct 6 14:36:22 2026
    From Newsgroup: rec.crafts.metalworking

    "Jim Wilkins" <muratlanne@gmail.com> writes:

    "Richard Smith" wrote in message news:m1tsn05zdv.fsf@void.com...

    So I was using metallurgy as I have gained with steels, castings,
    welding, etc., and the AI was identifying likenesses in geology as a
    path to show me the mechanisms of deposition of ores which we now mine.

    -----------------------------------------
    That would benefit from understanding oxidation states, reaction thermodynamics and solubility of metallic compounds.

    https://resourcecenter.kinetico.com/common-water-problems/iron-ferric-vs-ferrous/

    As with iron and carbon, when a melt cools dissolved impurities tend
    to become less soluble and crystallize out, forming grains of ore
    within the matrix of pure(r) rock. Slower cooling enables larger
    grains.

    Yes that similarity
    * "segregation" in castings
    * forming metalliferous-enriched "rejected" fluids in a solidifying
    magma pluton
    is one of the likenesses the AI was able to work with and tutor me.
    --- Synchronet 3.22a-Linux NewsLink 1.2
  • From Richard Smith@null@void.com to rec.crafts.metalworking on Tue Oct 6 14:43:25 2026
    From Newsgroup: rec.crafts.metalworking

    "Jim Wilkins" <muratlanne@gmail.com> writes:

    ... https://resourcecenter.kinetico.com/common-water-problems/iron-ferric-vs-ferrous/
    ...

    Sierra Nevada in Southern Spain - there are amazing springs on the South
    slope where the water comes out clear but tastes strongly of iron /
    minerals, and if stored in contact with air orange.

    "The sulphide belt" of Southern Iberia sweeps past/through this area...

    --- Synchronet 3.22a-Linux NewsLink 1.2
  • From Richard Smith@null@void.com to rec.crafts.metalworking on Tue Oct 6 14:49:46 2026
    From Newsgroup: rec.crafts.metalworking

    "Jim Wilkins" <muratlanne@gmail.com> writes:

    ...
    Diatoms are tiny marine organisms that extracted dissolved silica from
    the water to form their shells. Deposits of them are used in swimming
    pool filters. ...
    ...
    https://en.wikipedia.org/wiki/Diatomaceous_earth

    Diatomaceous earth also gives us "dynamite".
    The liquid nitroglycerine produced in vessels on top of sand-dunes and trickling down a lead-lined launder becomes fairly safe to handle once
    it has been mixed into diatomaceous earth.
    --- Synchronet 3.22a-Linux NewsLink 1.2
  • From Jim Wilkins@muratlanne@gmail.com to rec.crafts.metalworking on Tue Oct 6 12:42:49 2026
    From Newsgroup: rec.crafts.metalworking

    "Richard Smith" wrote in message news:m17bjv2bop.fsf@void.com...

    "Jim Wilkins" <muratlanne@gmail.com> writes:

    As with iron and carbon, when a melt cools dissolved impurities tend
    to become less soluble and crystallize out, forming grains of ore
    within the matrix of pure(r) rock. Slower cooling enables larger
    grains.

    Yes that similarity
    * "segregation" in castings
    * forming metalliferous-enriched "rejected" fluids in a solidifying
    magma pluton
    is one of the likenesses the AI was able to work with and tutor me.

    --------------------------------------- https://en.wikipedia.org/wiki/Eutectic_system

    Also:
    https://en.wikipedia.org/wiki/Freezing-point_depression

    The practical application is determining the weight of individual molecules
    by measuring the freezing point depression of a known weight, which gives
    the number of individual molecules. In many cases it's not obvious if molecules bind into pairs or not.

    --- Synchronet 3.22a-Linux NewsLink 1.2
  • From Jim Wilkins@muratlanne@gmail.com to rec.crafts.metalworking on Tue Oct 6 18:03:06 2026
    From Newsgroup: rec.crafts.metalworking

    "Richard Smith" wrote in message news:m1wlrv0whx.fsf@void.com...

    Diatomaceous earth also gives us "dynamite".
    The liquid nitroglycerine produced in vessels on top of sand-dunes and trickling down a lead-lined launder becomes fairly safe to handle once
    it has been mixed into diatomaceous earth.

    -------------------------------------------

    https://www.gutenberg.org/cache/epub/46039/pg46039-images.html
    Hudson was Hiram Maxim's brother.

    --- Synchronet 3.22a-Linux NewsLink 1.2