HN user
This is a great guide. However, the economics just do not work in my favor at all. Even if I were to spend $2k, I get much more flexibility of model intelligence and choice from a provider for $20/month.
This book equipped me with the right intuition and tools to visualize machine learning. I wish I was smart enough to hold it all together.
you're conflating a compute problem with a code quality problem.
Also, isn't it a great ad for Anthropic itself? One wonders
What is happening? I see multiple outages and CVEs is being reported on HN's front page. I've never seen these many security/incident related posts on HN's front page.
I don't like this proposal but engineers should not be shamed for doing their regular jobs. We all do it in some form or the others.
I don't think so. I think this is a common narrative in Hackernews when layoff news are shared. All the people I talk to in the industry positively confirm a boost in productivity. Its contribution to actual revenue could lag but it is present and confirmed by many.
Good luck, I'm sure you will find a great role!
It would be a reasonable deduction for someone who doesn't have the time or interest to understand the internals.
This is insane. I have to move to Codex now.
I agree, thanks for clearing it up.
Your argument is not meant to tackle my core claim, it is to poke pedantic holes. What a waste of my time.
You do realize that you need a human, a "SWE", to do the task that I just described? A computer can't do it.
A recent one is the RCA of a hang during PostgreSQL installation because of an unimplemented syscall (I work at a lab that deals with secure OS and sandboxes). If the search of the RCA was left to me, I would have spent 2-3 weeks sifting through the shared memory implementation within PostgeSQL but it only took me a night with the help of Opus 4.5.
To me, that's intelligence and a measurable direct benefit of the tool.
just a random token generator based on token frequency distributions with no real thought process
I'm not smart enough to reduce LLMs and the entire ai effort into such simple terms but I am smart enough to see the emergence of a new kind of intelligence even when it threatens the very foundations of the industry that I work for.
The only reason anyone uses a TPU is because they couldn't get the best GPUs.
I agree but one can admit their situation instead of outrightly rejecting the claims. My own mistake is to have become so hopelessly dependent on them.
I have a genuine dislike for all Meta products now. With time, their intentions have become much more clear and it was never to bring people closer or whatever.
Boris gaslighted us with all the quality related incidents for weeks not acknowledging these problems.
After a certain amount of context usage, I think I empirically see the stated issues with Top-K compression strategy. It doesn't catastrophically forget but nuances fade as I reach towards the tail end of my context limits.
I mean Anthropic clearly wins with the name (Mythos vs 'GPT-5.4-Cyber')
And the worse part is the company is gaslighting people when they report it
You're okay with sitting at the rear seat of a car while it drives you around the city though.
This was obviously a fictional thanksgiving dinner. Nobody is this geezed up about AI assistance.
It has nothing to do with the context window. Reasoning brought measured approaches grounded with actual tool calls. All of that short-circuits into a quick fix approach that is unlike Opus-4.5 or 4.6. Sonnet-4.5 used to do that. My context window is always < 200K.
I would have not believed your argument 3 months ago but I strongly suspect Anthropic actively engages in model quality throttling due to their compute constraints. Their recent deal for multi GWs worth of data center might help them correct their approach.
I suspect it's not that people do not see the progress, they fail to fully trust laws not truly backed by physics like the transistor laws. We empirically see that scaling works and continue to work.
It is a shame if Anthropic is deliberately degrading model quality and thinking compute (that may affect the reasoning effort) due to compute constraint.
Nope, there is a categorical degradation in quality of output, especially with medium to high effort thinking tasks.
The downtime forces me to relook at my utterly dependent relationship with agentic assistance. The inertia to begin engaging with my code is higher than it has ever been.