jm4

Probably network effects, integration, just about everyone knows how to use it, high switching cost, the fact that it’s mostly good enough, etc.

Even if you switch away from GitHub, you’re probably going to use it for something else anyway. There will be some dependency or something that’s still on there or you might use it for discovery. Then you have to spend time and money migrating to something else for what? Comparable features and interface? Maybe avoid a few minutes of downtime every month? Migrating a team might cost the equivalent of years of downtime. Or host it yourself to spend more of your own time and maybe get a worse result?

Really, what does downtime cost anyway when you are using a distributed version control system? It’s not like you don’t have access to the code. You might not have access to issues or CI for a few minutes. So what? It’s annoying, but is switching really worth it?

And what else is out there? Gitlab? Gitea? BTW, that’s the dumbest name ever for a product. I wouldn’t use it just so I don’t have to say that word. Are either of those enough of an improvement to justify the time? Or host Forgejo? Now you’re in the hosting business and only have yourself to blame when it inevitably goes down.

It’s easier to just use GitHub even if GitHub isn’t great.

I self host Forgejo on a cheap VPS with hourly borg backups to rsync.net. It's a great system and seems comparable to github, although I'm not a heavy user of actions on either platform. I do use the container registry though. It was a smooth transition and easy to install. I used the docker installation method. Forgejo is great if you don't mind hosting yourself.

Interesting. My experience has been similar. Opus 4.8 was awesome. Opus 5 feels a little off, although I can't put my finger on exactly what it is.

I probably have the same strix halo box as you. It's slow, although mostly tolerable, but even 128 GB isn't enough to run good models. Where that leaves us is giving money to somebody to get access to frontier models. I don't like any of those guys either, but some appear worse than others. FWIW, I mostly use Anthropic models. And CCP and their distilled models notwithstanding, at least they release open weight models at a lower cost. It's not like any American company can claim the higher ground these days anyway, especially when they trained their models on pirated content and trash our neighborhoods with their datacenters.

There's a big difference between quiet donations to a PAC - not that that's good either - and what Elon did. He literally paid for votes, likely in violation of the law. He poured more money into U.S. elections than anyone has before. He and Trump both made strange, cryptic statements about Elon's role in Pennsylvania with the voting machines that has caused people to reasonably wonder if they somehow manipulated the election. Whether he did or not, the innuendo alone is not ok. Then he did what he did in Germany. Don't even get me started on DOGE or his Starlink shenanigans in Ukraine.

That's before we even start talking about the models themselves. He claims to want "unbiased" models, but he very clearly has a distorted view of the world and has repeatedly demonstrated a desire and willingness to bend the world to his will. I don't want to use a model that is so obviously suspect. Not to mention, his models repeatedly produce racist, Nazi-like propaganda.

IMO, he is, at best, a clueless amateur masquerading as an expert and running into problems a more careful person manages to mostly avoid. At worst... well, you get the picture.

I have. He was using it due to philosophical reasons the same way many people have philosophical reasons for avoiding it. I don't know how many people are like that, but it's not exactly where you want to position your product if you're a business.

Personally - and I know I'm not alone with this sentiment based on comments I see on this site - I wouldn't touch Grok no matter how good or cheap it is. I don't trust Elon and I don't want to give another dollar to the world's richest person who turns around and uses the money to interfere with elections. The guy I know uses it for essentially the same reason I won't use it.

I remember having this on a Sidekick in like 2007.

100%. Personally, I hate Meta and I don't use their apps. There's so much bullshit and fake news on there it's not even funny. But my business uses FB and Insta because they deliver results.

I can get my content in front of tons of people for a pittance. I can upload a list of my best customers and run a lookalike campaign. I can run a re-targeting campaign on lapsed regulars. I can spend $16 to generate $1000. It's like fishing with dynamite.

Think about the type of person who uses FB. They don't care about ads. They don't care that it's fake content. They don't care that at least half the feed is AI slop. They just keep scrolling, clicking and buying.

Meta didn't create these people. They only got them all to gather in one place and then started selling access. There's a sucker born every minute and they grow up and join FB. This isn't a defense of Meta by any means. They are a horrible company and the world would be better off if business models like theirs were outlawed. The point is nothing is going to change with them unless there's legislative action.

How many of them do you think are actually there to ensure ethical and benevolent AI? They are like the Human Resources department for AI models. They don’t give a shit. Their job is to provide cover for the company and to make some psychopath’s shit circus look like a professional operation.

Purely anecdotal, but my experience has been that LLMs generate low quality python code. It's spaghetti code on par with what I've seen when companies I worked at tried offshoring development. It's basically what you get when you give bad or incomplete specs to a team of inexperienced programmers with poor development practices. It's usually good in small chunks, but it gets extremely sloppy as the scope of work increases and more decisions are introduced. Interestingly, I've seen LLMs generate good clojure code.

My guess is it comes down the the training data more than anything else, although I suspect functional languages will fare a little better. At least that's been my experience. There's undoubtedly a ton of python code in the training corpus and portions of it are of dubious quality. Niche functional languages likely have a smaller training corpus where a larger portion of it is better quality.

What makes you think it's small businesses? I would assume most businesses doing any reporting at all are using a payroll system. Mine handles the reporting automatically. The kinds of payroll systems used by small businesses can be set up in a day. If anything, small businesses that make even a small effort to stay on top of things have an easier time than big ones with complicated systems and processes.

I would argue that Microsoft had a solid run for some years in the middle where their software was pretty good quality. I think it started with Windows 2000 and certainly lasted through the XP era. Windows 7 was pretty decent too. Their operating system was straight up trash before win2k and has been in steady decline for some years now. It doesn't surprise me that younger people are only now (re-)discovering that Microsoft operating systems are bad.

Are you crazy? Do you realize how much use they allow under a subscription already? I’m going through $200-400 in tokens per day on a $200/month subscription. Nobody has ever said anything, throttled me, encouraged me to go to API pricing, etc.

I wish it didn’t go down so frequently, but it does. Still, I get an enormous amount of value. I realize they probably prioritize API users over subscribers and I’m ok with that. I use the API and openrouter for the things that need to be resilient to outages.

I would be embarrassed to try to ask for a refund or free credits. They already give credits far in excess of what you pay for and you have practically the entire month to use them outside of a couple hours of downtime.

My comment is partially in jest. Off the top of my head, I'm not aware of any particular policies that make brain worms more prevalent, although the person I was replying to mentioned prions and I believe that's a real concern. In general, this administration has favored a looser regulatory environment. Specifically, there have been policy changes directly linked to cyclosporiasis.

Last year CDC scaled back the Foodborne Diseases Active Surveillance Network from eight monitored pathogens to two. Cyclospora was one of the six dropped. The remaining two are Salmonella and E. coli. For what it's worth, there has also been an uptick in salmonella cases. Food policy experts have pushed back on the assertion that the policy change caused the outbreak, but concede that outbreak response programs are underfunded and the fact that cyclospora no longer being monitored makes it difficult to understand how it fits in with longer trends.

What the food policy experts say really points to the broader issue - an overall decrease in response and monitoring capability - and that's a bigger concern than playing the blame game on one particular outbreak.

It's reasonable to question how government agencies are making these decisions and one just yesterday on the peptide industry may be one of the best examples yet of how flawed the decision-making process is. The FDA's Pharmacy Compounding Advisory Committee (PCAC) met and voted in favor of adding several peptides to the list of substances compounding pharmacies are allowed to produce. Eight members of the panel had been recently appointed under RFK, including six who run clinics that offer peptides, increasing the size of the panel from 6 to 14. The votes on the peptides were 8-6.

So... back to my comic book villain theory. This is exactly how a comic book villain would do it and fool everyone with a veneer of legitimacy.

A guy gets a worm in his brain, somehow becomes secretary of health, loosens all the rules so the rest of us have a higher chance of getting a brain worm. This is comic book villain kind of stuff. Can we start a serious discussion about whether we might be living in a simulation with a troll teenager at the controls?

I don't know if this is serious or if it's a joke.

They're not a news publisher. Not exactly anyway. They are a huge conglomerate and Reuters is just one division - and a recent addition at that. TR is probably more known for legaltech, Westlaw and ancillary kind of stuff that ends up spilling over into other industries like tax and accounting. They also have a large government division, which presumably will take on this project. Their bread and butter is enterprise tech, consolidated public data and research tools. That's sort of where the news fits in. It's just another data source.

Probably because it consumes the same APIs originally developed for a web app and the same front end engineers likely created Claude Code. People assume Anthropic set out to create Code as a product and went through a normal design phase. I think it’s more likely it started as an internal tool or someone’s side project and blew up as a product when they released it to the public.

Am I the only one around here who’s sick and tired of the bitching and moaning on every post about how something was vibe coded or written by AI? Without fail, someone complains and it shoots up to the top of the comments. It’s gotten ridiculous and it’s off topic.

The easiest thing for you to do is just not engage with the post if you don’t like it. You people don’t need to pollute the comment section for anyone else who’s actually interested.

Just about everything is vibe coded or written with AI these days. Assume that’s the default. Comments pointing it out or complaining about it is just noise.

That assumes progress is linear, which it certainly is not. We’re also assuming Chinese companies are releasing their best stuff when our own government is dictating what American companies can and cannot release.

I don’t know for sure that they are weeks or months behind. I doubt anyone outside of 3 letter agencies knows that. The pace of AI is crazy fast and China is notoriously secretive. We could be comfortably ahead or China could have the top model by the end of the year.

I don’t think any of us have enough information to know what’s really going on, but I suspect it’s a very tight race.

70% is not unreasonable gross margin. What makes you think that’s a reason to ask that they lower prices? That 70% has to cover a lot of expenses.

The joke is that McDonald's spends hundreds of thousands of dollars to identify new locations - traffic studies, visibility, demographics, nearby traffic generators, site characteristics, drive-thru feasibility, etc. They have one of the most rigorous processes in the industry. Burger King's process is to open a location across the street.

That's not fair at all. I am not saying anything about too big to fail. My whole point is that most of the people who get punished by a corporate death penalty don't deserve it. And if you want to go around imposing that penalty then some things need to be structurally different to avoid all the collateral damage.

By all means, punish Zuckerberg and the executives who built the monster. Civil forfeiture is ok with me. Take their money, punish them, charge them with crimes if they broke the law. I don't care. I'm not going to advocate for them escaping accountability just because they are big. Force them to change their business model. Let them fail if they can't find a way to succeed legitimately. But understand that there's a difference between letting a business fail and killing it overnight.

That has far reaching unintended consequences. Thousands of people out of jobs, investors out of money, etc. We’re not just talking fat cat shareholders. There are institutional investors, pension plans, 401k’s. Killing a company like Meta has a financial impact on virtually everyone in the country when the people we’re actually trying to punish are the slimeball CEO and his inner circle of executives.

Our system can’t bear something like a corporate death penalty. Maybe that’s another problem in and of itself. You won’t ever catch me advocating for a company like Meta - they are among the worst - but the solution can’t be one with the potential to kick off a 2008-style financial crisis. The cure can't be as painful as the disease.

Good read. A studio can be both good and not good enough for Microsoft. They’ve never made a secret of the fact that they are in it for huge numbers. A studio can do well and still make little enough that it’s a management headache for Microsoft.

I agree with the mess part. I also feel like a reset is necessary and they are messing it up more right from the start. Fourteen layers of management? That's insane. But 3200 layoffs is a hell of a lot more than middle management cruft. Asha certainly has her hands full over there and the honeymoon period - if there even was one - is definitely over.

A few thoughts on the vibe coding… This is probably just one person and this project wouldn’t have seen the light of day if they weren’t able to vibe code it. A few years ago this would have to be a kickstarter that raised at least several hundred thousand- probably millions to have a shot at successfully getting off the ground. You’re talking software and hardware engineering, experts in multiple disciplines, a whole team of people pouring in many hours to develop a product, etc.

Vibe coding doesn’t always have to result in low quality. An experienced engineer with good systems design skills piloting an agent can be incredibly productive. Although I’m pretty rusty at writing code, I’m still good at systems design and I’m having success with coding agents.

Recently, I’ve built a system for myself because what I wanted didn’t exist. There’s no way I ever would have done it without AI. I wouldn’t be able to pull it off myself even with years of time and a budget to hire developers for my personal project is nonexistent. It’s the kind of thing I never would have thought to start prior to good coding agents.

My productivity has been insane. I feel like there’s 10 of me. The quality of output is shockingly good. I’m looking at this and it’s one of the most put together systems I’ve worked on at any point in my career. It’s beyond what I saw from much more senior developers than I and it’s beyond what I was ever capable of myself.

I get why people don’t like vibe coding. It does produce a lot of slop in the hands of someone unskilled in the use of their tools. It costs people their jobs. There are a hundred reasons not to like it. The flip side is we get cool projects like this one because a single person can build the thing they always wanted and never could until now.

For real. I've used 8B tokens in the past month and haven't hit my limits even once. In fact, I can't even get close except for the day I used Fable. I've barely stopped. Claude keeps reminding me to sleep.

It was already frustrating to use before. I wanted to review my own code for OWASP top 10 kind of stuff and it kept refusing. It repeatedly popped up scary warnings about how I was violating TOS. I had to go through quite a few iterations of that prompt. When I finally got it to work it burned through all my remaining usage on a single run.

I won’t even bother with it if they’ve made it even more frustrating. Instead, I’ve been using a combo of Opus 4.8, GLM 5.2 and DeepSeek v4 Pro. Then I have Opus synthesize and verify the reports from all 3 and make the fixes.

How many drivers with their own car live in a single family home? That 2/3 statistic is undoubtedly skewed by minors who don’t drive and live with their parents.