If the only accountability here is token input/output, after automating that employees would be messing around with no accountability either way.
If setting the actual goal you want to achieve as a manager, and then trusting employees to allocate their focus accordingly, you will absolutely have people faffing off (as likely can't be avoided), but at least those who don't will optimize towards what works according to their actual expertise at least.
Someone external to the curl team ran the test. If that third party found a severe CVE that they could use across all the global curl attack surface, and did not disclose it back to the curl team, the third party could keep using the exploit until discovered independently.
Sure, but isn't it a verdict on Mythos compared to other models?
If so, it would still follow. "Most software" isn't analyzed as much as curl, by either other tooling or other models, that might well find close to the same as Mythos did. As such, Mythos then isn't especially/particularly dangerous.
Coming up on 20 years here, and what I'll say is it gets easier, but increasingly sad is that the specific memories fade and only the more general memories remain.
It pains me that she never got to meet her grand kids, and vice versa.
Not GP, but: not confident. How confident would I be to avoid a (slightly lower entropy) UUID collision while also avoiding a clock desync landing on the exact same logged millisecond? Very, which is how confident I was about not encountering an UUID collision before this thread, so very++ I guess.
I can somewhat understand companies getting users depentant on their harnesses or workflow, but model vendors as in this deepseek case, I have absolutely 0 model loyalty when it's a simple config change away, and will always optimize for either capability or price (or whatever !/$ metric you can determine).
Depends what you’re doing. For example, Gemini is somehow still your only option if you need a model that can natively understand video and reference timestamps in its response.
As sibling said, Pi is great, and you can absolutely run it directly (there's even a plugin to use itself as a sub agent), but I mainly run it as a sub agent from other harnesses, for example running a more capable model in copilot, and then delegating simpler chunks to pi (using a cheaper model) as the sub agent. I've tried gas town and some others but never got into that way of working. I'm going to try opencode though as a less vendor specific harness than copilot/claude/gemini.
I've been using OpenCode for a few days now, I like it. It doesn't feel any "less heavy" than Claude Code (they're both massive piles of vibe-coded typescript) but for me it's essentially a 1:1 replacement for Claude Code.
Sidenote, I've been trying deepseek-v4-flash and I'm blown away. It's no Opus, but it's as cheap as tap water and punches far above its weight as a Flash model. I keep throwing tasks at it out of curiosity and it keeps solving them.
There are points in history where the productivity increases were more equitably spread, and we still got lasers, microwaves, MRI, mRNA, microchips, the internet, etc, etc, from national funding no less.
A milder version would be: People are paid based on the marginal cost of replacing them. But still, either of those is a far cry from claiming it's "the value they create".
It's not hard to show either. Imagine you're on an island and have been gored in the stomach by a wild animal, and a skilled surgeon can save your life. Most people would instantly agree that continued existence is of immense "value".
However after making that decision, you discover that you're on an island where almost everyone is a surgeon and the going-rate is much lower than you expected. Are we supposed to believe that your new knowledge somehow reaches backwards in time and retroactively changes the "value" of Not Dying? No, that'd be crazy. Valuation is never the same as price-point.
Elon Musk stopped producing value quite a while ago, nowadays all he does is tweet stupid stuff. Meanwhile his net worth has skyrocketed.
Everyone who's ever spent 6 months in big tech knows how easily compensation is divorced from value. Plenty of net-negative, work-creating behavior gets rewarded with big raises.
> Elon Musk stopped producing value quite a while ago
No evidence of that.
> Everyone who's ever spent 6 months in big tech knows how easily compensation is divorced from value.
You could always sell your yourself to a big tech company with your expertise in who are the value ones. A person with that skill would be very valuable to a company.
Sir, I understand that you did some cool, useful stuff in your career and got rewarded for it and that's a case of the system working, but if you can't see various sycophants and crooks getting just as rich or richer, while contributing nothing, then I can't see it for you.
Count his tweets over the last few years, WaPo did a big story on it recently, its like 70 tweets/day. Then divide his remaining attention over the number of companies he's running. Or look at the effectiveness of his management of DOGE which is the most public/transparent example we have.
I do think up until 2020 he was useful, there's a reason nobody else made electric cars or self landing rockets work. But most of his wealth accumulation has been since then.
Brother, both Twitter and Grok had to be folded into SpaceX to hide their abject failure while he was spending all day on social media. People change, he's not the same guy as the 2010s Elon.
And people demand the maximum pay they can get. It's the Law of Supply and Demand.
Consider this. You hire Bob for $10/hr, and he produces $100/hr in value. What's going to happen? Your competitor hires him away from for $20/hr. Then another competitor hires Bob away for $30/hr. This proceeds until Bob gets paid about $85/hr. The ROI of hiring Bob is somewhere around 15%.
There's a good reason why the vast bulk of the American workforce is paid much more than minimum wage.
You are correct that there is no friction-less market.
However, lack of information is called "risk", and risk strongly affects what you will pay for something. A low risk deal will cost you more than a high risk deal.
Risk is part of every transaction in a market economy, and is priced in.
The article starts out being wrong by using wages rather than total compensation. The latter is the complete pay package, which includes paid time off, health insurance benefits, retirement benefits, etc. These typically add about 40% to the wages/salaries.
If setting the actual goal you want to achieve as a manager, and then trusting employees to allocate their focus accordingly, you will absolutely have people faffing off (as likely can't be avoided), but at least those who don't will optimize towards what works according to their actual expertise at least.
reply