“If Meta wants…” My concern is that the only conceivable motivation Meta could have for investing money in such a project is making more money. If, in the process, Meta destroys the eco-structure of the Fediverse, so much the better—less competition, more money for them.

Oh, eww. A Meta/Twitter mashup? No. Just no.

I loathe Facebook, although my church uses it, so I’ll grudgingly check it once in a blue moon. Never quite got the whole Twitter concept. Used Twitter intensively only once or twice for simulcast commentary back when the Sharknado series was a thing. And found I could sometimes get faster customer service on (for instance) airline snafus, when company portals were unresponsive, and phone queues required waiting through 3 hours of drecklich Muzak.

Nope. Everyone makes mistakes. But you don’t go full Armageddon on the people whose blood, sweat & tears built you up from diddly, and then say “oopsie.” It don’t work like that, Spez. Have fun with your IPO.

Fascinating stuff, although I do not pretend to grok the maths.

On the one hand, seems to offer formulaic approach to creating effective interventions to curb online hate-speech. A more functional approach than hand-wringing, and more effective than “thoughts and prayers” platitudes following IRL consequences of hate-speech.

On the other hand, who decides what is undesirable hate speech? For instance, I could see both of the US abortion debate “sides” (yes, I know, there are more than two) claiming the other is wrong, is promoting hate or harm. Each side might feel itself justified seeking to intervene to prevent spread of an ideology.

Apparently, the substance and quality of thought Redditors post may be a disposable commodity. That is eyeballs on ads may be secondary to our function as generators of natural language as grist for training artificial intelligence—with third-party apps a civilian casualty in a bigger war for the almighty dollar.

According to https://www.msn.com/en-us/money/other/reddit-sparks-outrage-after-a-popular-app-developer-said-it-wants-him-to-pay-20-million-a-year-for-data-access/ar-AA1c06d9:

Part of the motivation for Reddit’s plan involves the surging popularity of artificial intelligence.

Large language models such as ChatGPT are developed using training data, which in many cases is sourced from content found across the internet. Reddit should not be expected to provide that data to “some of the largest companies in the world for free,” CEO Steve Huffman told the New York Times in a recent interview.