nixpulvis

You can't claim that the tech industry isn't innovating. You can say that innovation effort has shifted away from many things that still need it. The issue I have is that the companies and products are so large and complex that they are extremely difficult to compete with.

This is needed, but likely not enough yet. With the rise of Discord servers as community forums, we've lost the wealth of Q&A they used to provide. Being able to search for a user with the same issue as you and find it was no small part of what's let us all learn to program in the first place, as well as critical to the training data that powers the AIs we use to answer similar questions more quickly.

There will always be new questions, one that even AI cannot figure out alone. Things that require other environments or polled feedback. Making these channels indexable on the internet is critical to the progress of information.

Yes, this is the real continuation of that. I still don't agree with the deprecation, and hope the support will be continued.

> In Scion, every commit, rebase, or merge happens on the server, so there’s nothing to push...

I think this is the wrong way to solve problems and goes the wrong way toward solving real problems.

Maybe I'm a bad person for this, but I both love coding and designing abstractions, and I enjoy LLM use. It's like having my old college friends to argue with about whenever I want.

This has always been true to a large extent. Unless you're cleared to carve out your own space of a project, or you're working off an already detailed issue/feature request.

Doesn't pi just have 4 tools?

- read - write - edit - bash

I found this: https://arena.ai/blog/coding-agents-harness-tax

Seems to support my skepticism.

What is the harness doing that's not the model.

I've been assuming a harness is basically a set of tools and a TUI for passing text to the model and the model coming back with tool calls and user responses.

Are the tools really that complex and different between harnesses?

A coworker was trying to tell me that models perform better in their own agent harnesses. I use both `pi` and `omp` and I'm somewhat skeptical. I understand that the tool calls might be slightly different. But really how much impact on the model itself does the harness have?

I'm going to completely ignore the first part of this post because I'm not interested in arguing about how severe the issues with SHA-1 are. I think it's accepted that there are flaws.

So given that, I'm more interested in the arguments for why migrating to SHA-256 is problematic.

The biggest issue I see, after skimming over it, is the submodule breakage for new projects trying to link to old projects. This seems solvable frankly, but is the only serious issue I see. Everything else will be worked out as software is updated IMO.

I think LLMs providers pretty directly compete with content they scrape like Wikipedia and SO...

A reference to a value isn't the same as the value itself. I'm confused the point you're making here.

You could keep using the old compiler, no?

I firmly believe that tech companies being responsible for parenting our children is a deeply misguided response to a failing society and overworked parenting generations.

I benefitted greatly from early access to computers and the full scale of their capabilities. It was a constant issue in my family limiting screen time, and who knows what the exact rules should be for everyone. For me it was a dynamic ruleset I was actively involved in the development of.

I could make compelling arguments to my Mom, that will not be the case with Meta, or Apple, or Anthropic.

Accepting "OS level" age validation is surrendering to a system meant to protect the industry, not our kids future. And while there are serious conversations we should be having about how to control technology use, it should be about parental controls and sane defaults, not age validations and legally restricted access by providers.

I hate how even the most progressive-minded people on most fronts are falling into this trap because big tech has failed them and they are angry at it all.

Yea, that's sorta my point. We're not, or at least I'm not to the point I have a local model as good as Opus yet though.

It's not even that I'm slower at manual coding anymore, though I very well could be, it's that my expectation of my own progress has shifted. I'll be frustrated to only get a fraction of what I could have done.

Big difference though, once you know how to read and write, nobody can take that away from you.

The issue with AI is not its obvious value, it's our dependence on it. I worry what will happen to me when I am cut off for some reason.

Local models are going to be really important and hopefully computing costs don't make it too prohibitive.

I can't speak to the quality of the fuzzer since I haven't used it or looked at it thoroughly, but it does seem to cover a lot of ground on features and interesting concepts. I'll definitely be reading more into what you have here.

I'm not against physical keys at all. I have one on my keychain. I'm against making things hard to maintain ourselves. Being dependent on Apple and Google to agree before we can own our own keys is what bothers me.

Why is a password manager a "service" it should really just be a program and as many encrypted backups as you want. Maybe you pay for them to host those backups, but that's a choice you should be able to make.

How is this easier than just using a password manager? Why not standardize the password input mechanism so they reliably update, etc.

I feel like all of the security of passkeys could have been build in a compatible way with new standards and enhancements to existing password interactions.

Bug tech isn't a single entity. But I strongly believe a different group would have finished this obvious gap by now.

It's ao comically bad it proves the while thing is a joke unless you're trusting Apple or Google to sync and back them up for you. One cynical angle at why the backup is being slow-rolled is because the major players have an incentive to not do it. They want you dependent on them and locked into their ecosystem.

It's completely insane to treat a credential to an account as something that cannot be backed up. It implies there's another form of recovery, which likely means that key is only as secure as the other recovery options. And when it comes all the way back to the master key to your manager itself the loop falls somewhat apart.

It's a hard problem, but passkeys aren't ready for me yet.

It's not functionally equivalent, but it is a workaround, but requires doing it on every site (if they allow it) and it cannot be freely moved and re-backed up offline from the site.

The actual reason is people have many devices. I assume this is at least somewhat common, but I still avoid passkeys so IDK.

You're designing a system where we should just be able to backup our own keys if we want to.

Have we standardized a way to backup and export passkeys yet? Do websites commonly allow multiple passkeys to be registered?

I feel like if you're scared of rebasing, you don't actually understand git.

Not using big_space :(

It doesn't and I never said anything about preferences directly. There are however culture wars, and they cause problems throughout the industry and discipline.