God DAMNIT why did you repost this now I want grilled cheese

Additionally, this means you can backfill.

Rather than Grain and Spark etc. etc. needing to build out their own userbases, they can backfill from Bluesky and each other.
Different algorithms and different ways of surfacing content but still a lot more content than any platform could do individually

It links via DID right? No handle resolution during the outage

I've noticed that sometimes Claude will get stuck in local minima like this and you gotta kinda talk some sense into it or take away its toys so it snaps out of it

There is also the "feed of feeds" feed, which shows every post on Bluesky where someone linked a feed

Yep, there's a ton of feeds. You can even make your own feed if you wanted to.

(This is what that Attie thing that everyone got upset about actually does - let you make your own custom "for you" feed).

But I recommend browsing the feeds list and seeing what feeds are available

Yes, but small indie teams will never admit that they had LLM assistance for coding.

And it's impossible to prove one way or the other without looking directly at the source.

I predict that small teams will NOT use LLMs for assets but will quietly do so for code

I do think one day we will either see "Six Flags Knott's Berry Farm" or the Six Flags branding will be gone from basically every park that isn't Over Texas/Over Georgia

@pronounpicker.ripperoni.com helps a bit, you have to rely on other folks who are subscribed to the labeler though

But Bluesky isn't intentionally feeding your data to LLMs.

It's just a side effect of how this platform has always worked, and what makes it different than other social media websites

Bridgy Fed Bridgy Fed is a bridge between decentralized social networks like the fediverse, Bluesky, and web sites and blogs.

There are other systems that don't do this! For example, Mastodon. You can find a Mastodon server which has policies you agree with re: data sharing and set up shop there.

You can even use fed.brid.gy to bridge your Bluesky account to Mastodon (and vice-versa)

The reason this is open is so that if Bluesky gets bought by a billionaire, someone else can make a new Bluesky competitor and you'd be able to keep all your follows/posts/likes/etc.

The ONLY way that works is if that data is public. Otherwise they can hide your own data from you and lock you out

Bluesky isn't doing that explicitly; that's just a side effect of how the system is set up.

You have a PDS that is locked-open, meaning ANYONE (not just AI labs) can see ANYTHING.

Look, here's all your data: all your likes, blocks, follows, etc.
pdsls.dev/at://did:plc...

For programmers it is industry-shaping.

For everyone else it is basically useless.

But if you're a programmer, it lets you write better versions of the same code you would have written yourself (usually).

It isn't much faster but it can find (real) answers quicker than an average programmer can

Have you checked out @semble.so?

It uses your Bluesky account. You can use it to collect any arbitrary links, not just Bluesky posts
bsky.app/profile/semb...

This also worked in reverse - when someone was wearing business casual, you knew they had a meeting with studio execs that day

I used to work with folks who were ex-Disney Interactive.
Disney's dress policy is business casual. But that didn't apply to Disney Interactive, so they all wore the "programmer dress code" of a graphic T-shirt and jeans.
Everyone at the Disney Studios knew who the Disney Interactive folks were haha

Then at least admit your goal is to participate in a harassment campaign instead of trying to dance around it, because that is absolutely what you outlined here.

It may be uncomfortable for you to acknowledge that, but it is absolutely what you're participating in

Cool, so your goal is to harass them into... quitting?

And then what? What happens to this platform? What happens to your account?

You still have no right to be going in and harassing folks.

Take a second and put yourself into their shoes. Imagine every single post you make getting flooded with comments vaguely angry at you for things you don't really have control over.

Have some empathy and just think about how you'd react

You wanna do that shit, go to Mastodon and subtoot them from there.

At least then you're directing consequences (you leaving the platform) at the platform's leadership and not some random engineer who has no control over what T&S is doing or what the CEO's biases are

When the engineer you're talking to has NOTHING to do with the policy, and it's coming from multiple people across the site... it's harassment.

They did nothing wrong except work for a company whose products YOU ARE USING. Folks have no right to harass them over that

At a certain point I don't blame them for being outwardly hostile to people who they have to assume are coming at them from a place of malice

To be fair, they are constantly being harassed by users and have been for a while.

Like, it started with harassing engineers over the actions of the T&S team. Then it was harassing engineers over Jay's transphobia. Now it's because they're open about using LLMs (most programmers today use Claude)

It is VERY different than "I typed a thing into the box and everything magically appeared".

If you're coding that way, you're doing it wrong

Once the code is written, you as a human review it and think through it and make sure it's right.

Then you can ask a different LLM to double-check the code as well and make sure everything looks right.

Then you test your code and put it in for human code review. Then a second human signs off

Once the plan is done, then the implementation starts.

The nice thing is that as part of the plan you come up with ways to test your code. The LLM can make sure that your code is correct, because there IS a correct answer thanks to the tests you came up with

You can spend hours on this spec, just going back and forth about details.

You write a plan before making a single line of code. LLMs help with quickly finding "what is a function that does this" or "how does this work".

This plan is the fun part and requires a lot of thinking. And it takes time!

Have you used it to write code at all?

Look. I'm no fan of LLMs. But it isn't remotely equivalent, in a way that is difficult to explain to non-technical folks.
Basically - with LLMs there is a planning phase. You write out a spec and it looks at the codebase and makes verbal (not code) suggestions

Not that all platforms are respectful of listening to the license.

But many are, and those are the ones that get vetted for use by professional organizations because there's a legal risk to regurgitating e.g. GPL-licensed code.

So modern LLM coding platforms have safeguards to respect licenses