Hacker News: neversupervised

New comment by neversupervised in "SWE-bench Verified no longer measures frontier coding capabilities"

neversupervised — Sun, 26 Apr 2026 15:04:21 +0000

Terminal Bench is the future

New comment by neversupervised in "SWE-bench Verified no longer measures frontier coding capabilities"

neversupervised — Sun, 26 Apr 2026 15:04:09 +0000

But this is the good kind of goalpost moving

New comment by neversupervised in "Show HN: Terminal-Wrench, a dataset of 331 realistic hackable environments"

neversupervised — Wed, 15 Apr 2026 15:19:47 +0000

That paper focuses on breaking the harness, the same hack applies to all tasks. Here we are breaking tasks individually. If these were put on a different, more secure harness, most of the exploits would still work.

Show HN: Terminal-Wrench, a dataset of 331 realistic hackable environments

neversupervised — Wed, 15 Apr 2026 00:42:30 +0000

I want to share a new dataset of 331 reward-hackable environments. These are real environments used in Terminal Bench and adjacent benchmarks. I first got interested in this because, as a reviewer of Terminal Bench, I noticed a lot of our tasks were hackable. I also noticed that many contributors to the benchmark do so because it provides credibility when selling environments to labs. Hence, TBench tasks are, in my opinion, held to a higher quality standard than those being used today for RL. No one is spending hours manually reviewing the $1B in tasks being purchased by major labs. As far as I understand, while everyone knows environments are hackable, nobody has released hundreds of "realistic" environments.

Comments URL: https://news.ycombinator.com/item?id=47773298

Points: 6

# Comments: 2

New comment by neversupervised in "90% of CEOs Say AI Changed Nothing. The Other 10% Have a PR Team"

neversupervised — Tue, 14 Apr 2026 15:31:25 +0000

This is nonsense. I’m sorry. AI will completely upend the workplace and the economy. Whether that’s self evident today in the numbers in the way that we track those numbers, which is based on how things have historically worked, is not relevant. First principles thinking is enough.

C’mon. Stop wishing for a future that feels convenient. This is not the world in which we live. Everything will change. Let’s help people accept and react to that.Let’s stop with the comfort talking and false hope.

New comment by neversupervised in "How to Make a Good Terminal Bench Task"

neversupervised — Mon, 23 Mar 2026 18:41:15 +0000

I've been a contributor and reviewer for terminal bench since last August, and this post is about what I've learned designing and reviewing tasks. The guidance is broadly applicable to anyone building an agentic benchmark.I would love feedback from the HN community.

How to Make a Good Terminal Bench Task

neversupervised — Mon, 23 Mar 2026 18:39:47 +0000

Article URL: https://twitter.com/neversupervised/status/2035455298417430911

Comments URL: https://news.ycombinator.com/item?id=47493447

Points: 3

# Comments: 1

New comment by neversupervised in "Reports of code's death are greatly exaggerated"

neversupervised — Mon, 23 Mar 2026 14:34:14 +0000

Can you explain what you think will happen, actually? People at OpenAi and Anthropic aren’t longer coding by hand. Are you saying everyone changes their mind and goes back? Not gonna happen. You have to work around this new constrain.

New comment by neversupervised in "Reports of code's death are greatly exaggerated"

neversupervised — Sun, 22 Mar 2026 20:35:04 +0000

The author’s intuition is still backward calibrated, even though he talks about the future. He doesn’t have an intuition for the future. All code will be AI generated. There’s no way to compete with the AI. And whatever new downsides this brings will be solved in ways we aren’t fully anticipating. But the solution is not to walk back vibecoding. You have to be blind to believe not most code will be vibecoded very soon.

New comment by neversupervised in "Atlassian says it had right to fire engineer for suggesting CEO is 'rich jerk'"

neversupervised — Sun, 22 Mar 2026 16:19:53 +0000

There’s no reason a company should put up with enemies within. In rare instances a disgruntled employee might be able to make a positive contribution. In most cases, even if the employee has valid reasons, by the time they are disgruntled there’s no coming back. It’s best for everyone to move on.

New comment by neversupervised in "I'm OK being left behind, thanks"

neversupervised — Sat, 21 Mar 2026 14:27:58 +0000

The mistake is that 1 every N waves of hype are in fact monumental shifts and it makes sense to embrace as soon as possible. Also being early to the right thing can have massive implications in appreciating the shift before the general public, which is upstream from making smart resource allocation (investments, career choices). I have a friend that went to super early OpenAI as a designer. He has more equity than most AI researchers there and made a 0.001% amount of wealth. Being early does very much matter in the right conditions.

New comment by neversupervised in "Warranty Void If Regenerated"

neversupervised — Wed, 18 Mar 2026 22:35:40 +0000

I don't oppose reading AI generated content in principle, but because it's free to generate, I always am less likely to read super long prose that is AI generated. So the question is whether someone has taken the time to keep it as long as necessary but not longer. Or if there are ways to make it easier for me to commit to the experience, with a sort of TLDR

New comment by neversupervised in "UBI as a productivity dividend"

neversupervised — Sat, 14 Mar 2026 18:16:55 +0000

UBI will likely be necessary but that won’t appease society. Everyone wants to have a chance to climb the ladder. If it becomes self evident that humans can no longer have a meaningful impact on their outcome, there’ll be riots whether they have a roof and food or not.

New comment by neversupervised in "AI didn't simplify software engineering: It just made bad engineering easier"

neversupervised — Sat, 14 Mar 2026 16:05:02 +0000

The goal has nothing to do with you being employed. Your job security is a consequence of the ultimate goal to build AGI. And software development salaries and employment will be affected before getting there. In my opinion, we already past the SWE peak as far as yearly salary. Yes there are super devs working on AI making a lot of dough, but I consider that a particular specialty. On average the salary of a new grad SWE in the US is past its peak if you consider how many new grads can’t get a job.

New comment by neversupervised in "Coding after coders: The end of computer programming as we know it?"

neversupervised — Sat, 14 Mar 2026 05:34:54 +0000

It’s crazy how some people feel the ai and others don’t. But one group is wrong. It’s a matter of time before everyone feels the AI.

New comment by neversupervised in "Coding after coders: The end of computer programming as we know it?"

neversupervised — Sat, 14 Mar 2026 05:33:49 +0000

I think you’re a bit behind on your world view. Just because it’s inconvenient to you that non coders can now code, doesn’t make it untrue.

New comment by neversupervised in "Cloudflare crawl endpoint"

neversupervised — Wed, 11 Mar 2026 14:23:30 +0000

Tell more. Crawling is not a new idea. How did they abuse you?

New comment by neversupervised in "Yann LeCun raises $1B to build AI that understands the physical world"

neversupervised — Tue, 10 Mar 2026 15:16:34 +0000

Is it good? This will almost certainly fail. Not because Yann or Europe, but because these sort of hyper-hyped projects fail. SSI and Thinking Machines haven’t lived to the hype.

New comment by neversupervised in "Tell HN: I'm 60 years old. Claude Code has re-ignited a passion"

neversupervised — Sat, 07 Mar 2026 16:12:26 +0000

Interesting bifurcation between developers that get energized by AI coding and those that feel depressed. Only one side will come out on top, even if it’s for a limited time.

New comment by neversupervised in "Little Free Library"

neversupervised — Mon, 02 Mar 2026 06:04:31 +0000

I find these to be cute, romantic almost. But I have never found anything worth borrowing. I wonder what is the real impact in terms of additional books read. I do love the concept of spreading knowledge in the neighborhood. I'd be curious about other similar approaches.