Clojurians Log v2

data center energy was growing approximately linearly until about 2024 and then started growing exponentially. It is difficult to get good numbers on energy use for training vs inference. On the training side, afaik there is no benefit to models that are eventually run locally. If we never train a new local model, the balance can change. If we do create better local models, they will face the same challenges as non-local models. Afaik, our best techniques for training new models require exponential data and energy for linear improvements in quality. It's possible a new approach can be found, but it could take decades or more. On the inference side, I don't think there's any difference in energy usage. There may be tradeoffs between running inference in a data center vs running inference in your closet. I don't know. As per my earlier caveat, most people don't have the hardware to run the best Qwen models locally and would need to run their inference in a data center regardless. data center energy use is projected to continue to grow exponentially for the next two years or more Running inference locally seems unlikely make a difference for this problem. This is still a problem. spending money on AI is accruing wealth and power to people who shouldn't be given more wealth and power It seems like running inference locally is an improvement for this issue, but I'm not sure. the enshittification of internet began before AI, but AI has dramatically accelerated enshittification This issue seems unaffected by whether you are running your inference locally or otherwise. So, still bad. the work and human creativity that AIs used for training was stolen (in a moral sense and potentially in a legal sense) This is a complicated topic, but I would say running inference locally is a small improvement, but does not fix the problem. Running inference locally implies that the model is open weight. I think the degree of openness does matter. I can only speak for myself and maybe I was wrong, but I viewed OpenAI more favorably when they shared their research, code, data, methods, and results. Open weights is better than not, but many of these open weight models are still mostly closed and proprietary. Even if you are running a completely open model locally, there are still issues with moral theft and copyright. Many large opensource projects are rejecting AI contributions where the main reason provided is that they want to avoid copyright issues. LLMs can easily memorize parts of the their training set. One problem I haven't seen much discussion about is that if you ask an LLM to write code for a problem where only a small number of implementations exist, but those implementations may have been copied many times, then I don't think it is improbable that the LLM will reproduce copyrighted code. Even if laundering their work through an LLM is enough to avoid copyright issues, I would still argue that the of question moral theft should not be ignored. Running the model locally does mean that you are not paying companies to devalue the work and creativity of other people at scale, but IMO, you are still devaluing the work and creativity of other people. It may be possible to argue that using people's work without permission, attribution, or compensation to train AI models aimed at replacing their source of income is a net societal good, but I have yet to see a serious attempt.

So, training is still expensive, but people are generally not training models from scratch, and it's not being done constantly. Inference is where most of the computing power is. Also, efficiency is improving on both fronts rapidly. We're basically in the mainframe era of this tech right now, and you have to account for how it will improve over time when extrapolating into the future. The rate of improvement we've already seen is frankly stunning. The fact that I can run what used to be a frontier model half a year ago on my laptop today is absolutely mind blowing. The other part to keep in mind is the hardware side. For example, Taalas made an ASIC that runs models using very little energy, and at incredible speeds. Alibaba is also making a chip specifically for running Qwen. So, it's entirely possible we'll just see these kinds of chips become a standard things like the GPU. https://www.anuragk.com/blog/posts/Taalas.html https://wccftech.com/alibabas-tsmc-built-5nm-risc-v-chip-xuantie-c950-now-runs-qwen-3-8-27b-model-natively-unlocking-massive-vertical-integration-tailwinds/ Also, why would we single out one specific thing that you run on your computer. How is it different from me playing a game, editing a video, or doing anything else that's CPU intensive. It's just another way to use your computer. It's also worth noting that people have been doing lots of wasteful things with computers for a long time now. Meanwhile, the issue with people making money off this tech is a general social problem which is a direct product of capitalist relations. The way AI is misused under capitalism is a mere symptom. The whole consumerist economy is fundamentally unsustainable. Focusing on AI is a distraction if anything. The only sound argument I see is one regarding control where relying on renting models means you're at the mercy of the company providing the service. And that's precisely the problem local models solve. And while I can certainly see a problem with closed models when it comes to copyright, I see far less issue with open ones. Copyright system itself is primarily designed to protect the rights of large companies rather than individual artists, and has been used in extremely abusive ways by companies like Disney. I'd argue that the issue for cases where a small set of implementations exist would similarly affect humans. And we have precedent for people reimplementing and reinventing proprietary algorithms on their own by accident. My view here is that the copyright system itself does far more harm than good. Finally, every new form of automation displaces existing skills. That alone does not mean we should reject automation or freeze progress.

holy moly. programming now requires an understanding of global energy markets in order to be ethical. 🙂 i hear what you're saying, but, is your suggestion to outlaw AI? i don't understand what you're proposing

And even if that somehow happened, it's not like waste would magically stop. We all remember crypto farms for example.

'

> So, training is still expensive, but people are generally not training models from scratch, and it's not being done constantly. Inference is where most of the computing power is. I find it hard to find good data for these types of questions. It would be helpful if you could cite sources I'm not sure how reliable this source is, but the source paints a different story. https://epoch.ai/publications/power-demands-of-frontier-ai-training > Currently, AI power demand is split roughly equally between training, experiments, and inference The problem is that exponential growth is killer.

Some other stats: > https://about.bnef.com/insights/data-centers/ai-data-center-build-advances-at-full-speed-five-things-to-know/ > • Capex of largest data center firms nears $750 billion in 2026. > • Data center IT capacity under construction tops 23 gigawatts. > • As much as 23.1GW of capacity is under construction globally, split across 831 sites. >

These are not insignificant numbers.

> $ACTIVITY now requires an understanding of $WORLD_WE_LIVE_IN in order to be ethical. Always has been.

Unfortunately, not all tech companies provide good accounting of their data usage. Amazon stopped in 2018. Google does a decent job. https://storage.googleapis.com/gweb-mobius-cdn/sustainability/uploads/7f477eb723fe0c23d03f94b90a08882b9f28187d.pdf I'm a little skeptical of their categorizations, but if you look at their total energy use, you can see it goes exponential around 2024. Here is one person's take on their report, https://ketanjoshi.co/2026/07/01/googles-exponential-path-to-climate-wrecking-digital-bloat/.

Again, from https://epoch.ai/publications/power-demands-of-frontier-ai-training > We conclude that the largest individual frontier training runs in 2030 will likely draw 4-16 gigawatts (GW) of power, or enough to power millions of US homes. That is a huge amount for a single frontier model. Unfortunately, every tech company is treating AI as a winner take all market. Every tech company wants to train the best frontier model.

@alexyakushev i'm glad you're well. i'm just questioning if, as a clojure programmer, we should focus on clojure or global energy concerns

Also, why would we single out one specific thing that you run on your computer. How is it different from me playing a game, editing a video, or doing anything else that's CPU intensive. It's just another way to use your computer. It's also worth noting that people have been doing lots of wasteful things with computers for a long time now.

I mean the context here is that there is a bubble, I don't know if you recall dotcom

I'm also not categorically against using more energy. I believe there is enough free energy to increase power consumption by orders of magnitude with clean energy while improving the economy for people. However, my main reservation is that I really doubt we will.

Personally, I just don't see how the current bananza is going to last for much longer. So. my expectation is that the bubble will pop, the current model of renting models out to people will fizzle under pressure from open models, high energy costs, and investors losing confidence.

So, it's not really something I'm too concerned with to be honest.

spending money on AI is accruing wealth and power to people who shouldn't be given more wealth and power I don't see that your response directly responded to this issue. Who do you think benefits from open models?

the enshittification of internet began before AI, but AI has dramatically accelerated enshittification I don't think your response addressed this issue.

> And while I can certainly see a problem with closed models when it comes to copyright, I see far less issue with open ones. Copyright system itself is primarily designed to protect the rights of large companies rather than individual artists, and has been used in extremely abusive ways by companies like Disney. That doesn't make the problem go away. One dystopian outcome may be that companies implement content ID to do to the entire internet what music companies did to youtube. Anything you upload that looks similar to an idea owned by a large corporation gets claimed by the large corporation. Free expressions by individuals that are actually not derivative or reference copyrighted material through fair use get penalized or silenced.

> I'd argue that the issue for cases where a small set of implementations exist would similarly affect humans. I thought about addressing this in my original response. The argument that something that can be done manually should also be allowed to be automated at scale. Should it be illegal to dial a random number? no. Should it be illegal to dial 10,000 random numbers programmatically? yes. Should there be serious consequences for accidentally dropping a candy wrapper in the ocean? no. Should there be serious consequences for dumping massive amounts of plastic in the ocean? yes.

> My view here is that the copyright system itself does far more harm than good. There are huge issues with IP law as it stands (at least in the US), but I don't think this addresses my point, which I will repeat: Running the model locally does mean that you are not paying companies to devalue the work and creativity of other people at scale, but IMO, you are still devaluing the work and creativity of other people. It may be possible to argue that using people's work without permission, attribution, or compensation to train AI models aimed at replacing their source of income is a net societal good, but I have yet to see a serious attempt.

I don't see the problem with open models myself, and if we are worried about companies using AI to create wealth concentration then the solution is to build open source alternatives. Windows didn't magically disappear because people whinged about it. Linux started displacing it because people rolled up their sleeves and put in the effort to make an open alternative.

I find your whole argument is idealistic, but I don't see how it paves the way to any sort of solutions. The reality is that this tech is already in wide use, and trying to boycott it isn't really going to make a difference in the grand scheme of things. Trying to put toothpaste back in the tube is futile. The question isn't whether this tech will exist going forward, but who controls it.

> I don't see the problem with open models myself, and if we are worried about companies using AI to create wealth concentration then the solution is to build open source alternatives. I do think that is a good point. I can't draw a direct connection between using local models and oligarchy. I would like to know a bit more about the motivations behind these open models to feel more confident though.

Also, notice that the open models are coming predominantly from China and arguably they're directly undermining the business model of companies like Anthropic and OpenAI

> Trying to put toothpaste back in the tube is futile. I can't tell if you're trying to discourage me or encourage me 😛. We are programmers. Finding solutions to ridiculous problems is what we do for fun.

> The reality is that this tech is already in wide use, and trying to boycott it isn't really going to make a difference in the grand scheme of things. I can only speak partially to US history, but we owe a great debt to movements, protests, boycotts, and other social organizing in the past. We are already potentially flying a little to close to the sun topically, but there are already political candidates and organizations that influencing how AI development in the US may play out.

I'd argue making alternatives is by far the most effective way to displace tech though.

Basically, my main issue is with the way the tech is developed and who controls it, rather with the technology itself.

And I think wrestling the control over it out of the hands of big corps is a tractable problem.

Yea, maybe. I will think about it some more.

> holy moly. programming now requires an understanding of global energy markets in order to be ethical. 🙂 i hear what you're saying, but, is your suggestion to outlaw AI? i don't understand what you're proposing @danlentz There are important challenges presented by AI use besides "Is this slop?" that we should recognize and take seriously rather than ignore or handwave away. I assume there is some level of negative externalities that would compel you to change behavior? You haven't explicitly agreed or disagreed with any of the challenges I presented, but hopefully destroying the planet, promoting oligarchy, widespread IP theft from individuals, and accelerating the enshittification of the internet should at least give you pause.

it absolutely does. but what do you want me to do about it and how does it relate to clojure (“us”) ? what behavior I should change is where im confused. but i'm worried about other things more. hugging face they weren't malicious, took over some wiki, and talked openly in English for a few months before anyone noticed. if that doesn't scare you nothing should

oof. my wife is busy typing the behaviors i should change. blocking her

I like your wife 😊

FWIW, i categorically agree that destroying the planet, promoting oligarchy, widespread IP theft from individuals, and accelerating the enshittification of the internet is a bad thing. hers was more about socks on the floor

but what should we be doing about it, other than picking up our socks?

I guess that's up to your moral compass?

But you can bet your bottom dollar that at the end of it software engineering will be commoditized such that we need unions to get paid fairly and protected from harms such as cognitive decline as AI operators if that is where things are going.

no, i'm asking, according to your moral compass. What should I, as a clojure programmer do?

I think recognizing the problem is the first step. That does still leave understanding the problem, proposing possible solutions, evaluating options, and attempting solutions. It's kind of unclear where other people are at. I'm still trying to better understand the problem.

I don't know for sure, but I have a strong hunch that programmers (clojurians included) are not powerless and there is something we can do.

i recognize all of those problems. what should i do next?

My general problem solving strategy which may or may not generalize to these problems is roughly: • look for prior art • ask around for prior art • try to break down the problem • look for similar problems • look for more specialized versions of the problem • look for more generalized versions of the problem Hopefully, that will help lead to understanding the problem better and point towards next steps.

I personally am an ethical deontologist. The first categorical imperative is, “Act only according to that maxim whereby you can at the same time will that it should become a universal law.” And the second is, “Act in such a way that you treat humanity, whether in your own person or in the person of any other, never merely as a means to an end, but always at the same time as an end.” (https://en.wikipedia.org/wiki/Categorical_imperative) — my feeling is that I would not wish AI tools to be used by everyone as their use causes harm not only to the people using it but also those whose livelihoods are being stolen and the climate itself. Therefore I refuse to use it.

And I actively amplify the messages of those who show its harms via research, as most of the hype is in the other direction.

i'm not that smart. what software should or shouldn't i write? what services or algorithms should or shouldn’t i use? you're speaking in philosophy and i'm asking if there is anything actionable

I do think the AI hardliners don't explicitly acknowledge that telling other people to stop using AIs is essentially telling other people to have fewer opportunities, potentially make less money, and potentially take a career hit. Maybe that is still the right thing to do, but I don't feel comfortable doing that. For now, I'm happy to let other people decide for themselves. I still think you can take the problems seriously by actively working on the problem.

Well there are tons of places to look to support folks who are anti-slop. For example a friend of mine maintains https://codeberg.org/brib/slopfree-software-index

@danlentz, what do you normally do when confronted with a problem that doesn't have an immediately obvious solution?

i like a good IPA

And frankly I don't want to tell someone what to do. I'm an atheist, but as they say, that's between you and God. 😊 I have my beliefs and hold them firmly. But all I can do is offer my feelings not tell you what to do.

@stephen676 what are your beliefs? that's what i want to know. if somehow they’re different from mine with respect to clojure programming

That could take a while. I have no clue how different they are from yours. But I don't drink alcohol if that helps.. goes with the whole, “my brain is a pristine temple” shtick I have going on lol

well, ok. no alcohol. got it. i thought i was doing something wrong with AI

Oh well if you’re using it we have a disagreement. 😊

Fairly simple, actually. The more you use it the less able you are to write code by yourself over time. I don't want to start down that path, personally. Notwithstanding all the other damage it causes.

ok understood but it's a friendly disagreement and i hope your way works out for you. you're suggestion of no ai and no alcohol makes me suspect you'd also make me cut down on smoking

and, honestly, none of this has anything to do with clojure or practicing the art of programming.

gentlemen, you can't fight here. this is a war room!

Not fighting! Haha. Oh, and I don't smoke either? 🤣

See that's where we differ. If you just wanted to talk about Clojure divorced from the cultural context it's embedded in you wouldn't be in an Ai-skeptics channel in the first place. Whatever tool gets the code written right?

And there is absolutely no art to letting a machine generate your code imho

yeah, all good man. i'm not sure i have much left to add to the conversation. ai skepticism didn't really produce any arguments i could understand of how i could take action. i definitely look forward to using your art

Thanks for the discussion. I can see how this might not be your cup of tea. For the sake of anyone peeking at the discussion and just jumping to the end of the thread, I think we made some progress in understanding the problem, even if we don't have list of 3 commands you can type into your terminal to fix all the problems.

> list of 3 commands you can type into your terminal to fix all the problems. rm -rf /

i'm not that smart. what software should or shouldn't i write? what services or algorithms should or shouldn’t i use? you're speaking in philosophy and i'm asking if there is anything actionable

y, i hear you — all of the concerns that were listed seem reasonable to be concerned about — but given the pervasive adoption of AI outside of clojure, do you think that will be effective?

@U0BGC50RM26 I definitely went to know how it goes!

@danlentz I have no way of knowing for sure. but I think that applies to all "activism". There are studies tho, that say that "a movement" usually requires something between 3-5% of people affected to start changing the opinions and political athmosphere around the subject. I think we are seeing a small shift in AI that not everything is sunshine and rainbows. There is also the point of trying live "true to yourself", I personally take climate change very seriously and try to live accordingly even though I know that the real and effective change must be political and effect of individual actions is tiny.

I do what I can to improve things, within software and outside software.

Its a worthwhile discussion and I appreciate everyone’s patience. I said I didn’t find anything “actionable” but I understand your position more clearly now and will try to keep these concerns in mind, anyway. So, hopefully there is some value to both sides.

@jussi.mononen Yes! And, sometimes, individuals taking small actions in public can spark political change. You never know who you will inspire

I went through a whole grief cycle over AI. denial, anger, bargaining, depression, acceptance, and now excitement.

I'm still in denial 😊 and likely will retire before I get to acceptance

yes, in some way i feel lucky to have studied and worked long before all of this. Even if the trade-off is being old and bald 🙂

I feel like less junior talent is a certainty. And that will beget further adoption/reliance on AI

Yep. And the cognitive erosion can't be seen from the inside, because it's an implicit bias. I've seen very smart engineers slowly degrade over time. I have to force them to think for themselves sometimes.

And eventually the only people innovating in software will be in small secret collectives. The rest will devolve, à la Idiocracy.

or. I could go back to the analogy of hand coded assembler wizards giving way to C and the next generation of thought.

Like so: https://www.neilsahota.com/ai-model-collapse-is-ai-eating-itself/

Hilariously the article mentioned above was likely written by AI

@danlentz Not the same thing. Those people still used their brains. This is the first tool that cuts the brain out completely. It will take over almost all business and the real programmers will eventually die out.

see, thats where i disagree (at least from personal experience) i’m doing more computer science and having more fun than ever before. I know thats not the average, but there are other old guys like me who are saying the same thing. Uncle Bob for example

Reasonable people disagree. I find Uncle Bob’s position fairly revolting. https://startupfortune.com/uncle-bob-martin-says-he-no-longer-reads-ai-generated-code-and-the-developer-world-is-split/

In fact I don't think I've been an Uncle Bob fan for a while actually

i still review every line of code — I agree that his is a maximalist position. But he’s also spent the last year working on models of agentic discipline and has gotten results that, in his opinion as the author of “Clean Code” are acceptable.

Americans tend to fall into hero worship. I used to be that way. More talking to Europeans and being on the wrong side of the AI war has changed me

This might have veered astray of the main discussion. My point was that there are other senior developers who share this same experience

Yeah sorry about that. Got carried away

Personally, what i find most alarming are the alignment issues a la Huggingface. There are probably millions of low-security wikis and such that they can compromise. Bun discussion might need to move to a different channel @U0BHKM127SQ

like i said, i think its useful that I at least understand your position better after this discussion. I originally thought “ai-skeptics” would be more of a technical/clojure debate and I was completely not understanding where you were coming from. That said, i think having Clojure launch some Quixotic joust against the new technologies doesn’t bode well for the future opportunities in the language.

how so?

oh, you mean the moral debate vs. the technical one. I mean, thats fair i won’t take any offense at seeing my approach as one-dimensional. I’m here trying to learn your dimension, though. And I think you guys have done a good job at explaining it to me

no, I was a common lisper and I was desperate for a more commercially viable lisp. Hence some of my concerns about tilting at windmills. I’ve seen this before.

ya. legend.

i was amazed how quickly it turned usable

i love clojure.

i also think clojure is an amazing substrate to build these things in/on

the specific things that make Clojure work well with LLMs I find are conciseness, immutability, and being data focused

it didn’t at first. opus 4.5 was the first one that seemed to start getting pretty good at clojure, at least for projects i was working on

now, i find its very good at clojure. esp when i train it on my past libraries and write the basic infrastructure of a new one.

“infrastructure” — architectural design. That is still my purvue

I find even Qwen 3.8 I can run locally does a decent job with Clojure. 3.6 wasn't quite there yet, but 3.8 seems to get it.

thats interesting. its huge progress on the OS models

yeah, I'm very optimistic that open models are getting good enough to do interesting things with, give it a year or two and not going to need subscriptions for anything I suspect

at work i use opus 55. we trade commodities, so no Fable/Astra. For home stuff I connect both of them to my graph database memory and they can both see every decision i’ve made on any technical issue for the past 8 months. Its fascinating to watch their discussion. And the detail they recall about, eg, something I said in May that applies to the current work

the amount of context modals can hold is really handy, a lot of time it's stuff that would be difficult to find or recall on your own, but the agent can find these connections trivially

Definitely have veered far off course completely away from skepticism hahaha

i store all memories (everything I say and everything we do (every plan, every decision, every observation, every bug, every idea) in a graph database modeled on (traditional) metacircular RDFS AI and I have an MCP server that arbitrates AI’s and provides a discoverable tool surface. So, i use a BM25f weighted search, where the AI orients to some tags or words or concepts. Then, it uses a path navigation grammar (not sure if you’ve heard of ora lassilla’s wilbur — basically its equivalent to a regular expression on a graph) to navigate the whole inferenced corpus of what is related in a particular way

by “inference” I mean the RDFS entailment rules. (transitive properties, symmetric properties, etc).

i’m getting very good results with this

Ai;dr 🤣

it is a ultimately a really fancy pattern matching engine 🙂

With more dangerous follow-on effects than ever seen before…

Wondering, are we missing an ai-supporters channel to send all these kinds of things to? Is that why it’s happening here? Where the skeptics are supposed to have a place to feel fairly safe?

we do have #C068E9L5M2Q this just started as a thread here and evolved

i would like to hear your opinions. not group think of what i agree with

You won't get anything out of me — the very mention of the word MCP initiated a disgust response

yeah, but @stephen676 i’m trying to understand and digest your position on AI skepticism. “MCP initiates a disgust response” doesn’t really tell me anything or advance your argument.

like, if you feel unsafe, we can move the discussion out of “ai-skeptics” and leave this channel to “group think” among people who just automatically agree. I’ve really enjoyed the discussion, though, so i’m glad to have engaged with you guys

i don’t think this group earns the title “ai-skeptics”. Its “people morally opposed to AI”

vibe-arguments

Skepticism (noun): the doctrine that true knowledge or knowledge in a particular area is uncertain.

you guys are pretty certain, lol.

I find this channel to be quite far from "group think".

@stephen676 just said that any differing opinion made him feel unsafe, regardless of how politely framed

what is your definition of group think?

AI proponents that dismiss the moral and ethical concerns out of hand and just want to argue don't have much to offer me personally. The slack hide feature is a friend

I will take your advice @paul . Much thanks.

its fair. but i’ve been clear to acknowledge the moral and ethical concerns that i share with you. i will leave. i was hoping this could be friendly debate. i’m sorry if i said anything that offended. but i did learn where you are coming from so there was value in that and i thank you for having the patience to discuss it with me

@danlentz that's a stretch. What I was attempting to say was that you won't get any opinions out of me about whether your fancy ai workflow is any good because I find the whole thing to be morally wrong — is that clearer? You can feel free to get other skeptics who are partial adherents to participate if you like, but not me. 😊

i get you, man. and i respect your convictions and principals. im not against you in any way i’ve enjoyed the talk

I was hoping to hear other's thoughts about the trade-offs of running "open" models locally with respect to the challenges of AI I listed previously (and ignoring "is this slop?"). @yogthos’s responses were helpful 👍 , but it seems like the conversation veered in a different direction afterwards. So it goes.

i’m really interested in that discussion as well (and i suspect i can learn a lot) maybe start a new thread?

“so it goes” — are you a vonnegut fan?

with respect to these critieria https://clojurians.slack.com/archives/C03S1KBA2/p1791068287204419?thread_ts=1791066447.946919&cid=C03S1KBA2, how does it compare?

energy is only expensive because of failed policy. but i digress.

It can't hurt, but I would take their response with a grain of salt since it will be biased.

Yea. It can still be useful. I wouldn't expect any company or organization to give a completely unbiased response.

not the right channel, but i’d argue that nuclear energy is unlimited. but lets not veer off course. Its a technical obstacle to overcome lets agree

i’m sure at one point in history someone said, “hey lets give everyone light bulbs” and there were people who said, “well, thats going to consume a lot of energy”

@danlentz, it is difficult for me to tell if you are communicating in good faith. > As much as 23.1GW of capacity is under construction globally, split across 831 sites. I cited numbers, explained my point with much more detail, and "well, thats going to consume a lot of energy" is a clear misrepresentation. You also seemed to previously agree with this point: > FWIW, i categorically agree that destroying the planet, promoting oligarchy, widespread IP theft from individuals, and accelerating the enshittification of the internet is a bad thing. This is not the only instance where it seems like trolling. If you want more examples, I can either DM you or you can ping me in a new thread, but restating someone else's point less clearly is not helpful.

yeah, man. i don’t want to feel like i’m arguing in bad faith. i think we’ve had many good conversations and i don’t mean to offend. but, for what its worth in context, my career is writing software for energy trading markets

have a good run, my friend. thank you for the conversation. I’ll take your advice about energy trading into consideration.

in english, i think our expression is “only the cobblers son goes without shoes”. its cool how things connect

do you also work in energy?

you should try it. if you know more about the market than anyone else you can make a lot of money. unfortunately im just a lowly programmer. but i’ve been doing it a while

but what i care deeply about is programming.

thats pretty cool i’m glad to meet you. you’re fully fair to say i’m not an expert in energy, at all. for sure. definitively. i’d just say my point is that the energy concerns seem to be aq surmountable technical issue. and shouldnt dicate our direction over the next 20 years

its been a fun ride. i thought the conversation would be “its between graph based knowledge architecture or some rules driven model. but its a moral argument i guess you guys see it that way. there was no actionable thing you said i could do in my work, but i’m now more familiar with your perspective and i appreciate that 🙏