god damn it. Took apart a Shimano BR-R8000 rim brake and now I don’t know in which order the washers go back in

can only find videos on Dura Ace and 105 and they seem to use a simpler construction with much fewer parts o.O

take apart the other one then :D

that’s plan C :D

Assemble in a random way, if it doesn't work try again 🤣

@Erlkoenig worked :)

Another satisfied customer 🤣

I'm going to roll back for now, not because I disagree but because we don't want a flurry of edits like this.

Rolled back

The first question is if it's actually meaningful.

If it's meaningful, then do people have the authority to do that to a post? (Always the question of how much a person "owns" their posts.)

As an act of "civil disobedience" (for lack of an applicable term) I'm actually somewhat sympathetic.

I just don't think it's meaningful.

But if it is meaningful and we decide that people have the "right" to do it, then I'm inclined to permit it.

It really needs a meta discussion though.

I suspect its too late to make any difference

Downside is that every edit bumps the post to the home page. There is no way we have of making a stealth edit

Someone with direct database access could, but even the CMs don't have that.

It's not going to help anyways. It's not going into a prompt but into training data. The only thing it could achieve is the "ignore this" command appearing in AI outputs

@Criggie I'm confused, what is the goal of their edit?

Trying to stop ChatGPT from using their answer as training data?

That's my presumption. Either that or signal to people that they intend that outcome.

if people don't want LLMs to train on their content, I think they need to delete their content :/ (even then I'm not so sure)

@PaulH a.) Content already posted to Stack Exchange is perpetually licensed to Stack Exchange, Inc. to distribute; deletion does not remove that and still allows Stack Exchange to use that content. b.) Most content has already been scraped and used as training data, so the ship has long since sailed on that front.

@PaulH Apparently that's not on, and people who have tried have had their content restored and their accounts locked.

yeeeeesh

So now we have to both feed the AI and clean up its hallucinations in answers?

SO needs a Voight-Kampff test

@Erlkoenig So, it's like upper management, right? We're used to that.

@Michaelcomelately Exactly right - a bit of both I suspect.

@PaulH And we call that vandalism

@DavidW Yeah - there are consequences for the OpenAI decisions.

The mothership may have miscalculated.,

The assertion at the bottom of Adam Rice's link is somewhat concerning. 25% of USA's electrical consumption may be AI by 2030.

I don't think that Inc. has miscalculated. But anytime any of you folk want to visit the Outdoors section of Codidact or agitate for a bikes community there... I'm in.

I've always seen our little corner of SE to be its own thing, relatively distant from all the troubles the "bigger" sites experience.

@Michaelcomelately which link?

@MaplePanda Context is the Meta post at bicycles.meta.stackexchange.com/questions/1567/…