MeetingsBrowser

The docs feel AI generated.

There are a lot of words related to language design like tree walk interpreter, algebraic effects, and semantic analyzer. But no real content about the language design, motivation, or what distinguishes it from other languages.

Also strange to start of the language introduction with an entire page dedicated to comments and how white space is handled.

If this was written by a human, I how it is taken as constructive criticism to consider what key points you want to convey before people bounce off of a page.

But does Goodhearts law apply to things that were never a good measure to start with :)

I have heard from multiple people at smaller to medium sized orgs where token usage and AI adoption are a central part of performance reviews.

Anecdotal, but appears to be common among other comments in the thread.

For companies doing this there is no 'justify the expenditure'. Employees are being praised for high expenditure, regardless of actual outcome.

Leadership see the problem as 'people resisting AI'. Embracing AI is seen as the solution, and token usage is seen as the measure of success.

sort of a Michael Scott approach of asking people to quit so they don't have to fire anyone.

Someone should gather all the creatively worded layoff announcements and put them into a museum.

Aside, none of these announcements even attempt to make sense.

GitLab's TAM is exploding, demand is through the roof, LLM tooling is making each IC more productive, and to capatalize on this moment GitLab is

... "transparently restructuring" by asking employees to quit so they don't have to lay off as many...

The repo for this theme is no longer available

> The page you are trying to reach either does not exist, has been removed or you are not authorized to view it.

This feels shallow, but one of the biggest things keeping me from using zed is the lack of a good default dark theme.

The defaults all feel very low contrast, gray on gray that makes the experience feel dull and off putting to me, even if the editor itself is great.

I don't think so.

The people leaving LLM replies are paying minimum $20/month for LLM access, and probably more in practice.

A one time $10 fee is not a deterrent.

The "AI implementation" step in my workflow includes separate agents dedicated to testing and reviewing changes. The self feedback loop catches a lot of errors and mistakes, but it rarely produces working code in one go.

In my experience, the generated code handles the happy path, but isn't great about edge cases or writing clean code, even with explicit instruction in the initial prompt.

We usually end up doing multiple iterations with what claude/codex output, pointing out issues, asking for changes, etc.

As a software engineer who also hires other software engineers, I’m curious about the disconnect in our experiences.

I do systems programming. Before AI feature development roughly went like, design, implement, test, review with some back edges and a lot of time spent in test and review.

AI has made the implementation part much faster, at the cost of even more time spent testing and reviewing, though still an improvement overall.

We do not see the weeks to days improvement though. The bottleneck before was testing and reviewing, and they are even bigger bottlenecks now.

What kind of work do you do, and what kind of workflow were you using before and after AI to benefit so much?

There are definitely some tasks that AI has made 10x or 100x faster, but not the tasks that make up my day to day.

For me, there may be one thing I do every few months that AI is really good at.

The overwhelming majority of the work I do, LLM tooling is just ok at. Definitely faster overall, but with lots of human planning, hand holding and course correction.

I would estimate LLMs make me, on average 50% more productive , which is huge! But from my experience I cannot believe anyone is experiencing a 8h/5m multiple productivity boost overall

The craziest thing about AI is you can just try it yourself and check if the claims are true.

I use Claude code and codex daily. They have become an integral part of my workflow.

There is no task that takes me a day that they can complete in five minutes.

Even with the lightning fast progress being made, it looks like LLMs are a decade or more away from being that good.

If AI can do your job for you, you should be the first to know. Just try it and see!

Your username is con artist lol

He knowingly made up lies about the Sandy Hook parents for personal gain, and continued to do so on a regular basis for over a decade.

The case didn’t even make it to trial because he refused to turn over documents, likely because they would prove guilt.

Free speech protects your right to say your opinion, but it does not protect you from willingly causing harm to others for personal gain.

I have a seat in my company's enterprise plan and never gave a phone number.

You need to onboard with your personal phone number at work?

Why not with email?

xAI built the Colossus data center in 122 days (just the physical construction time).

Colossus initially had ~200k GPUs. 100B buys you ~1 million high end GPUs running 24/7 for a year at AWS retail prices.

Going from a company with no experience building and operating datacenters to a company with 100B worth of compute is a multi-decade high risk goal.

I disagree. Easily reviewing and combining multiple streams of parallel work is more valuable than ever.

I don’t dismiss them because I disagree, but because they measure something tangential to the point.

There is data on average software salaries and job counts.

Salaries have not grown since 2021. Inflation is high in the same time period, meaning pay is substantially down over the last five years.

Actual jobs are harder to measure but various sources all show swings of +\-5% each year.

Salaries are down 10%, and best case scenario jobs are up 5% in that same time frame.

Job postings are not jobs. A company can layoff 10k and post jobs for 1k and have job postings be up.

Projections are useless. The past does not predict the future.

BLS is useless for the same reason, amplified.

To be fair, the status page tends to lag by thirty minutes to an hour.

And I’ve experienced 500 error codes in Claude code that lasted more than thirty minutes but less than an hour that never showed up on the status page as an outage

Gifts come with some implied responsibility from the giver and a niche hobby project is different from a package manager.

Take it to the extreme. What if I write a library, put an OSS license on it, advertise it, and then bundle malware in the release.

Am I fault for including malicious code, or are the users who downloaded it entitled for expecting the code, that I asked them to use, will not harm them.

I would argue the burden is mostly on the user for smaller niche projects, but mostly on the developer for large, heavily advertised, critical infrastructure projects.

It is not entitlement to expect operating systems, package managers, browsers, etc to be following good practices.

I pay for the lowest plan. I used to struggle to hit my quota.

Now a single question consistently uses around 15% of my quota

Once you advertise and ask people use your software in production, you have an obligation to make sure it is somewhat safe.

If you actively advertise and give away free food, there is a baseline assumption that you are at least cooking the food in sanitary conditions.

If people get sick after eating the food you gave them, you can’t just shrug and say it was free.

> So, I have opinions about criticism of crates.io for supply-chain attacks.

I also strongly disagree with most of the criticisms of crates.io, but…

We are owed supply chain security. The moment a group says “use our stuff for critical projects” they take on some baseline level of responsibility for making things secure.

You cannot offer a taxi service in a car that is not fit for the road, and then just shrug when it crashes a people get hurt.

The good news is the people behind crates.io and the Rust ecosystem care about security. They have given conference talks about what they are doing behind the scenes. Features like Trusted Publishing are a huge step in the right direction.

As far as I can tell, the issue is not with the crates.io team, but funding and incentives as a whole. We all rely on critical infrastructure like package managers, but not many are willing to fund big security improving features.

I would fly commercial

> It is effectively security theatre.

I disagree. Security is always a trade-off.

Owning, auditing, and maintaining your entire supply chain stack is more secure than pinning hashes, but it is not practical for most projects.

Pinning your hashes is more secure than not pinning, and is close to free.

At the end of the day, the line of trust is drawn somewhere (do you audit the actions provided by GitHub?). It is not possible to write and release software without trusting some third party at some stage.

The important part is recognizing where your "points of trust" are, and making a conscious decision about what is worth doing yourself.

Exactly. Deterministic artifacts alone are not necessarily more secure and are tangential to a lot of what is being described in the blog post.

The blog is mostly focused on hardening the CI/CD pipeline.