top256

Did you find tune a model?

That distinction is really useful. My critique is aimed at how often “inevitability talk” blurs those two levels together. It’s one thing to say “networks need a lingua franca,” it’s another to say “TCP/IP was inevitable.” When people collapse the concept into the specific implementation, that’s when the rhetoric becomes persuasive but misleading.

I agree that physical and economic constraints matter, but the point I’m making is that cost curves themselves are contingent on human choices. ICs didn’t suddenly become cheap on their own; they became cheap because governments and corporations poured billions into scaling them. That’s not denying the underlying advantage, but showing how rhetoric of inevitability erases the political and economic decisions that actually drive those curves.

"They're describing reality" - No, they're making predictions about the future. If AGI requires 30 years of compute improvements as you say, then it's not reality, it's a forecast contingent on those 30 years of development continuing unimpeded.

As for "oppressive regime", we already do this for nuclear and biotech, and most people find it quite palatable! Nuclear materials are tightly controlled globally. Cloning humans is illegal almost everywhere. We've had the knowledge for both for decades, yet basement nukes and basement human clones aren't happening.

I'm not saying we should make it illegal, I'm just saying there are more gray areas than what's generally accounted for.

thanks a lot! I appreciate it

Thanks for your thoughtful response. I think there's a misunderstanding (maybe my text wasn't clear. If so please point out where so I can fix it).

We actually agree: even if the probability of successful coordination is only 10%, accepting inevitability makes it 0%. That difference matters enormously given the stakes. My argument isn't "coordination is definitely possible" but rather "believing it's impossible guarantees failure." When tech leaders say "AGI is inevitable," they're not describing reality; they're shaping it by discouraging attempts to coordinate. Human cloning hasn't happened because we maintain active resistance despite technical feasibility.

You're asking for credible paths with P > 0. I'm saying: knowing P with certainty is impossible, so accepting P = 1 narratives makes alternative paths invisible. The path emerges through trial and error, not before it.

Great example. Thanks a lot! Can I add it my essay?

Thanks for the link to Tom Renner's article. I've read it and agree with it.

I wrote this essay to convince "unbelievers" that's why I tried to be as rigorous as I could

You're absolutely right that predictions exist on a probability spectrum. I focused on the binary for rhetorical clarity, but you're correct that some predictions are more likely than others.

That being said, it doesn't change my point about agency. Your prediction #2 (LLMs impacting the economy) seems "almost inevitable" precisely because thousands of people are actively working to make it true. If everyone stopped tomorrow - if OpenAI, Anthropic, Google, etc., all pivoted to other projects - would it still be inevitable?

The appearance of inevitability comes from observing massive coordinated human effort toward a goal, then mistaking that effort for natural law. It's like watching a thousand people pushing a boulder uphill and concluding "that boulder inevitably goes up."

Thanks for engaging. I think you meant it's obvious rather than shallow? If so, yes I agree but a lot of my friends disagree so I wrote this as rigorously as possible to explain why this is obvious.

That being said, regarding game theory and coordination: Wars DO end when people change the game's parameters. WWI's Christmas Truce happened despite every incentive against it. Catalonia chose not to pursue independence despite having voted for it. The Montreal Protocol solved ozone depletion despite classic tragedy of the commons dynamics.

My point isn't that coordination is easy - it's that treating it as impossible becomes self-fulfilling. When tech leaders invoke inevitability via game theory, they're choosing to accept those constraints rather than working to change them.

Thank you for the comment and engaging with my thinking.

You're using hindsight to define inevitability, which is exactly the circular reasoning my essay critiques. "It happened widely, therefore it was inevitable" isn't a useful framework, it's survivorship bias.

Using your IC example: they became cheap because of massive government investment in the space program and military procurement, not natural law. The Apollo Guidance Computer alone drove early IC demand. Different policy choices = different outcome.

Personal computing almost died multiple times. Xerox PARC had it all in 1973 but management killed it. IBM thought the market was ~5 computers total. The Homebrew Computer Club was nearly shut down for copyright infringement. Any of these inflection points going differently changes history.

Your "no moat" observation is telling - you're really describing business strategy (technologies that spread can't be monopolized) not philosophical inevitability. But even that's questionable: TCP/IP could have lost to OSI, the Web to Gopher or AOL's walled garden.

The counterfactual test: if these were truly inevitable, we'd see simultaneous independent invention everywhere. Instead we see: singular inventors, path dependence, and technologies that almost weren't (or actually weren't: where's our supersonic passenger travel?).

Calling only successes "inevitable" while ignoring what didn't happen or was actively prevented (nuclear proliferation, human cloning, various chemicals/drugs) demonstrates the selection bias in this thinking.

yes that's exactly what I was trying to convey. Technology is a human byproduct so we are collectively in charge.

thanks!

I wrote an essay critiquing Silicon Valley’s obsession with calling technologies “inevitable.” I argue that inevitability isn’t a fact but a rhetorical move that erases agency and responsibility. What if we treated predictions not as destiny, but as challenges?

Yes I did and it solves tons of problems of coding agents

Also it wasn't done by a human but by my tool (the code in the repo is decompiled bytecode)

Great idea. Which agent to use?

I tried with opus and o3 but I had to copy/paste the code and I wasn't sure it was the best way.

I tried 10 prompts and the simplest was the best (probably due to the code being simplistic)

Yes exactly and that was the hard part (extract and verify the invariants). Still it's surprising because llm needs to be able to do that for any complex code.

What you wrote is great can I copy/paste it in the blog post? (Referring you of course)

It's a great idea but yeah the evaluator (especially in this case) seems hard to build. I'll think about this because it's a great idea

Oh good point which one should I use? I ran it alos with o3 and claude sonnet but the results were similar or worst.(Some are in the repo)

Yes but the analysis it's going to run is probabilist vs the one I'm running is closer to smt

Good point I should have mentioned. I was so "in it" that I didn't realize it was a valid question.

This thing outperforms llvm (on o3 and O2) and the JVM too. I didn't explain it because I wrote it in a previous blog post but that's a fair point.

And it's not that obvious because the previous example was handled better by chat gpt while llvm couldn't handle it.

Thank you for this idea.

TL;DR A loop that carries any pure scalar state can be strip-mined across p threads by having each thread privately replay ≤ p(p-1)/2 “warm-up” updates before its first public iteration. No closed-form skip-ahead, no speculation, and a few extra machine instructions in code-gen.

I agree and I think there's a market for faster computers!

What you describe IMHO is the lack of hardware progress.

Nowadays,hardware performance is increasing at 10% per generation https://www.cpubenchmark.net/singleThread.html

I think that's something fundamental that has changed. Hardware performance is increasing at 10% per generation https://www.cpubenchmark.net/singleThread.html

Do you see your hardware getting significantly better at each gen?

of course on https://deviantabstraction.com

I built it; happy to discuss!

That's exactly what I was looking for. I'll give it a try

Thank you for working on that!

Hey HN! Creator of RefRem here. We are two geeks working from Europe (Marcus is a cofounder of Pond5.com). We’re hoping to get feedback from any of you who are JVM users running serverless workloads.

Cold startup is a huge problem routinely taking above 10-15 seconds with spikes in RAM usage as high as 240MB -- both these problems discourage adoption of Java serverless.

RefRem (https://refrem.com) automatically converts your jar files into a bundle ready to run in an AWS Lambda. During our tests RefRem reduced RAM x5, and cold startups showed drops from 10 seconds to 100 ms.

Please sign up, and test it out. We are eager to see what the community will make of it.

Marcus and Nicolas