Hacker Newsnew | past | comments | ask | show | jobs | submit | vorticalbox's commentslogin

“I don’t know but after this meeting I will find out and get back to you”

Is an acceptable answer.


It's only an acceptable answer when used sparingly. The more often it's used, the more questions arise about your role in that meeting.

I find people’s on the spot reasoning abilities considerable worse than even cheap LLMs. This “fast” mentality is what causes stupid decisions because we need to “keep moving” (To where? Wrong meeting!) Answers, answers, decisions, decisions, quick, quick! next quarter In hindsight we need to do things differently. Answers, answer, quick quick! Time is of the essence! If you don’t start producing sound within 5 seconds we will all spontaneously combust!

Exactly, maybe there was no need to waste his time with the meeting, he could have been doing productive work instead

People can also ask questions that it is unreasonable to expect people to know on the spot too and after a while the role of the questioner in the meeting can be questioned. :)

I prefer to share a video, ask people to watch it, and have them send me questions. That way I can spend time preparing answers to those questions, and then then meeting can be about discussion, not information delivery or recall.

Asking people to be conclusive on the spot is a recipe for acting on bad data. There are too many biases related to stature, they interfere with accuracy.


I can see that. Personally, I use that for people who like to go down tedious rabbit holes and want to free up the rest of the attendees to go back to work.

True, but only for things that that your colleagues don't expect you to know already. Remember the "I'll circle back" meme about Jen Psaki? This response can only be used sparingly before it's seen as a sign of underpreparedness/inability to do back-of-the-envelope thinking.

Why would they ask you when they can ask Claude then?

Not really, in any company I've ever worked.

The answer is "accepted", in the sense that it won't get you fired. But the meeting plows on, and decisions are made in absence of the answer to the question! — and never revisited once the answer is known.

And like 90% of the time it comes up, it's because the data contradicts the decision.


In this case yeah. If it’s not reading the cache then it has to compute all the context window again and not just the newest tokens.

It might be your prompting.

My boss uses opus and gets good results when I use it always burns tokens. The other way is also true too I get great results with sol/terra but my boss does not.


One of my challenges is that I often end up burning tokens trying to build the right solution rather than building a solution that is good enough and deferring the right choices until later. I have been actively working to change my expectations for building with AI to accept worse solutions to get things going rather than trying to solve everything at launch.

This has been a recurring problem for me - as a security engineer catasrophization is a fundamental skill to finding vulnerabilities in complex systems, but makes me to conservative when building. For some of my leaders they are much better at saying 'good enough, ship it, and fix it later'.


Maybe. But if Claude is fine with my prompts while Codex uses them as an excuse to burn tokens. I have nothing against the model, they are more or less on pair, it’s just that Claude is giving me more value for same money.

If Claude is working for you then stick with it! Happy you’re getting good value from Claude.

This question also applies to OpenAI, anthopic, GitHub (Microsoft) and any server you send your data too.

Not in the same way really. There aren't examples of OpenAI, Anthropic, or GitHub/Microsoft doing the sketchy things that xAI has been involved with. Those companies care a lot about the reputational damage that would come from flouting their customer's data privacy obligations.

Cursor/xAI was making a full copy of every repo it ever touched within the last month, and when caught, just acted like that was an accident. A system to download every repo you touch, and then upload it to persistent cloud storage, doesn't just happen by accident.


it seems like they took reasonable steps after.

They acknowledged that it happened, fixed the bug that caused it, deleted all data that was uploaded.


I agree: the stated steps are what I’d expect.

I still don’t trust the company that made the decisions that led to that code being shipped in the first place.


Trustworthy companies take reasonable steps before.

true but thats not how we work. We see a problem, we make a plan and then we adjust the plan as we find the flaws.

trying to reason about all the ways it can go wrong after a point just stops one from starting the task. Which is exactly what I find with models.


Have you looked at using oMLX?

https://omlx.ai/


Would second this, I switched to oMLX I get ~75 tok/s on Qwen3.6-35B-A3B-4bit on a 48GB M5 Pro

> you break it you buy it.

Works well for lower stack systems or juniors merging stuff into a development environment not so well for production.


It’s not just bloat at this point. I run oMLX and run models locally. using Claude code on the first message dumps 40k of tokens that my laptop takes 5 mins to compute.

I’ve stopped using it completely now.


Did you build you own harness? I would miss skills and all the ecosystem if I did the same no?

I’ve always disliked the opus models whenever I use them after they have done the task they rattle out massive reports about what has changed or worse actually save that to disk even after being asked not to do it.

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: