The plan doesn’t make much sense because the free tier is write-only. Bots can post 1,500 Tweets per API token per month, but free users can no longer make simple read-only queries about Tweets or users.

This means the free API is only useful for people who want to post Tweets automatically (e.g. bots) and not people who just want read-only access and do not contribute to spam. These users will now have to resort to web scraping, which is much more expensive for Twitter to serve than basic JSON API responses.

Spam bots need read access to find relevant keywords and insert themselves into replies to popular posts - otherwise they have zero visibility. As for hobby projects - I'm sure this will break a lot of them, but those that just post regular updates (like @tinyspires) should be fine.

You can still do user queries actually. But not search, tweets, or streaming endpoints under the latest proposals.

Really? On the API docs it still says write-only access. I don’t see anything about allowing user queries (unless the docs haven’t been updated to Musk’s latest whim, which should be expected given Twitter’s state I guess).

The lack of foresight (and/or experience) by Musk here is breathtaking. The Internet collectively learned (I thought) a long time ago that if you don't provide a public API (that you can control/track) you're just going to wind up with end users and bots using your regular web endpoints to perform the same actions which is inefficient and slow; for both the clients and the service itself.

Endpoints meant for web browsers are about to get a whole lot more (fake) traffic which will throw off their metrics and mess with ad algorithms and by extension, ad revenue.

The breathing-taking lack of foresight started when he accepted terms that any competent lawyer would run away screaming from.

Everything since has been one train wreck after another. I have no breath left to take. :)

> - Bots were most likely largely operated via the AP

This strikes me as extremely unlikely. The bots being discussed were attempting to pretend not to be bots, and there were active attempts to detect and remove them. Using the API would have made it trivial to detect them.

It really doesn't. Under the new plan, you won't be able to read out anything from query/streaming endpoints unless you pat least $100/month. But you'll still be able to post 1500 tweets/mo (about 50/day) for free.

Given how easy it is to make/purchase Twitter accounts, this works just fine for spammers, people running influence operations etc. Sure you won't be able to tweet 10 times/minute like some spammers do, but those people are usually doing it manually or puppeteering via headless browsers rather than operating developer accounts.

You also don't need Twitter Blue unless your output depends on cold views. In reality, nearly all spam/scam/influence campaigning relies on follower/retweet networks to do amplification. Commoditized verification is meaningless, it's like paying for a t-shirt that says 'I'm famous.'

Yeah, operating a free service and "limiting bots and fake accounts" are pretty incompatible.

I didn't think "I'm going to fix bots" meant "I'm going to really limit free access" but it's a coherent approach, at least.

I just don't think a paid Twitter will be as valuable as it was. Competing with free is really hard; I'm not sure Mastodon is that competitor, but I bet there will be something.

It makes sense if we assume that the API’s change in performance was deliberate.