supriyo-biswas

Set noprocrast=99999 on your settings page.

With laws like Chatcontrol and all, other jurisdictions are not necessarily any better.

In fact, we're increasingly seeing a desire to build the "backdoor" directly into the software, e.g. mandatory age verification, client-side scanning, etc.

I must wonder whether it's their watermarking initiative[1] forcing certain logit choices to produce watermarked text that ultimately causing the model to behave in a dumb manner.

[1] https://support.claude.com/en/articles/16266773-how-claude-m...

> Building a SAT solver into the package manager seems to be a solution in search of a problem.

I can't tell you which ones off the top of my head, but I'm sure a number of package managers do use constraint solvers to find dependencies matching the constraints.

> I also remember things like "have one backup in one place, another in another place" since before this mnemonic

I've heard variations of this all the way back to some 1990s vocational IT courses in the country I live, and it wasn't even technologically advanced at the time. It's hard to have any sympathy for someone who misses this concept in today's age.

I'm only being forced to use it at $WORK since some people overran their Cursor bill, so everyone gets Cursor Auto enabled by default which routes to Grok 4.5.

The trick here is to have connections that follow and like content that is worth following, ever since connecting with some peers in the database space, my feed is filled with things like new research, C++ performance optimizations, and the like, instead of management fables and feel good stories.

In the age of uv (and pipenv before that), should we even go through the trouble of building Python manually?

I can't help but think that this is a submarine[1] paid for by OpenAI or Anthropic as a way of garnering public sentiment for AI regulation.

[1] https://www.paulgraham.com/submarine.html

It’s about being able to change models mid-task. For example, I want to be able to plan using Fable but implement the plan using Sonnet, and that won’t work if this is implemented.

qbittorrent has one, https://wiki.archlinux.org/title/QBittorrent (search “web interface”)

> Not saying that OP is living their life wrong or anything, to each their own of course, but I prefer working from home precisely because of how much more _life_ I get to live as a result.

They have to find a job that allows them to do that, and such jobs are increasingly difficult to find.

Where I'm based, even though it might be the smallest startup that could save a good bit by getting rid of their office costs if they gave people WFH, they won't.

> I get exhausted reading LLM prose.

While my advice is specific to learning about codebases, the way I do it is to have it generate mock data and put it in the local development environment, and give me some exploratory commands, and then ask away. It's a machine after all, so I don't have to read its preceding prose to understand whether it did tell me something, it can just repeat it however many times I ask it, and the hands on commands etc. give me something to actually try and implement.

I have tried this with Codex, and if you ask it to implement a project based on an algorithm or design in a given repo, it always puts an attribution link saying that part of it was adapted from that repo. Unfortunately I think we’re being duped here.

Is it really that complicated? I personally use direnv with nix to have per-project dependency versions installed automatically, and I’ve never had too much trouble. I know that the flake.nix files can become somewhat complicated, but mine have been pretty simple thus far.

I mean we've had Matt Mullenweg from Wordpress make various below-the-belt comments about their competitors and other companies, something tells me that they're not the only CEO who acts in such a childish manner.

I'm sure saner heads like their legal people would prevail at some point, but until then...

In the past "RTFM" or "LGMTFY" have been responses, so why not something like, well, "I'm not aware of the top of my head, you afaik <brief explanation>. Also feel free to ask Claude on the technical details :)"

If I may ask; what's your experience doing this with smaller models like gpt-5.x mini/luna on the lowest reasoning setting? Does it work well?

I've been thinking of doing the same since the models will often take design decisions that I am not a fan of, or make changes in a way that is easily understood by me, which leads to tech debt and being unable to anticipate what it will take to deliver a certain feature (e.g. whether we should build on top or build from a clean slate, etc.)

> It seems pretty unlikely they'd ever okay this intentionally as some sort of marketing. > Members of OpenAI's leadership team were actually in the middle of a Twitter spat with HuggingFace employees/open model advocates

The point of these campaigns is to eventually invoke some sort of response from the government, such as banning open models, which OpenAI (and Anthropic) stand to benefit from.

I’m assuming, somehow, “Dario and Amanda” has become a glitch token.

Farming is a necessary activity and in most countries, they are also a politically influential group.

The case for data centers in the age of AI is weaker, and the only effects for the average person are noise, pollution, higher electricity costs, higher water costs, for the ultimate "benefit" of losing your job.

I had a Logitech ergonomic mouse and by default, the two extra buttons corresponded to forward/back, and I had gotten quite used to them for navigating codebases. It's a shame the mouse didn't last long, though...

As for the squiggler, put a wrist watch or something underneath it?

I would hope that they really didn't move in the direction of having the model memorize image IDs, as that could be easily replaced with an embedding-based image search tool call.

Although, it is possible that it was unintentional as the model was RLHF'ed into memorizing images.

> LLM generated code is a ticking time bomb, and is going to poison projects.

Given the recent Anthropic ruling and OAI settlements, and the general stance of courts to look the other way when it comes to LLM use, I think an exemption for LLM-based software will be handwaved into existence (to my regret.)

I could do this in the beginning of 2025. Unfortunately, given the over-proactiveness of model these days, I find that they end up inferring what my original request was and implementing it anyway, and then writing unit tests, etc. even when all I asked for was to wire up a few components, as an example.

Well, it could make a note in its memory and reference it later. I don't know if such a statement from the LLM would actually cause a write to the memory in all cases, though.

> Perhaps being informed of your obligations when you create an account is insufficient

I'm sure that if we were talking about a corporate entity, then people would have retorted with how "no one reads the ToS" and "most ToSes are unenforceable", and guess what? In a world where significant AI assistance is becoming common, finding out which project is AI generated or even knowing the level of AI assistance is well nigh impossible and could be very well unenforceable.

Edit: It has come to my attention that Codeberg is a registered association[1]. It would be "fun" to see them argue about how they may argue that an AI assisted project has "harmed their reputation".

[1] https://en.wikipedia.org/wiki/Registered_association_(German...

I think then we should ban these sorts of posts about the allegation of distillation, since being able to post the story but then warning accounts with comments about the hypocrisy, is not the correct way to go about it.

Honestly, it wouldn't surprise me if they just found evidence of distillation once in 2025 against some Chinese AI lab, and they've been lying about the rest to create a narrative.

Everyone’s saying that it looks like Codex but TBH it just looks like all the AI chat websites on the side and some frequent actions on the top.