77 points by fniephaus 3 months ago
More details here: https://blog.twitter.com/developer/en_us/topics/tools/2018/e...
>”Site Streams, User Streams, and legacy Direct Message endpoints, originally slated for retirement on June 19th 2018, will be deprecated on Wednesday August 16, 2018 which provides 3 months from today’s release of the Account Activity API for migration.”
This is a huge “fuck you developers” from Twitter. They should have offered a year from the GA date to port applications. They should be sending emails to all applications using streaming APIs.
I like the webhook model, I think it’s great. I think the beta they had was effectively useless - for instance, the Docs were (still are?) entirely incorrect and basically told you completely wrong information with copy/paste errors everywhere. 0 QA. https://twittercommunity.com/t/account-activity-api-all-acti... you couldn’t register more than one webhook and were locked into a “env-beta” URL. Again, the Docs were wrong here.
I haven’t had a chance to review this since GA but I will now. I’m not hopeful that it’s any better given the track record of Twitter to deliver. If they did knock it out of the park I’ll be the first to praise them... but 3 months is still not enough time.
> This is a huge “fuck you developers” from Twitter.
Well, they are doing it again. If developers didn't learn anything from the previous time why should we care?
They announced the deprecation date back in December.
Yet they didn't give access to replacements until today.
What should we do?
Twitter offered a base service that was great and they are now constantly removing features (no chronological timeline, no access to polls for 3rd party apps, now no streaming for 3rd party apps). A few days ago on HN we were talking about Mastodon but I don't see it replacing a Twitter as it's a platform, not a single entity that I can use.
The best scenario I can imagine would be to see the teams behind Twitterific and Tweetbot creating a Mastodon server and offering me to switch in the app in the coming months. Twitter is screwing us because we are, as users of those apps, the 1%. And I'm quite sure that the people I follow are also part of the same single percent.
To The Iconfactory and Tapbots: make me pay again, ask me to pay a subscription, I don't care. But screw Twitter, they treat you (and thus us) like garbage, make me switch to something offering me a better experience and that is guaranteed to last. Is posting tweets from my account through your apps still free? Then offer me the possibility to send tweets when I send messages on your new platform + a weekly message to send to my few followers that they should also switch.
Twitter is gonna apparently kill you at some point anyway, letting your users see a stream ordered by a shitty IA and with 1/3 of ads. So please, let's fight instead of slowly dying.
There's still an option for a chronological timeline. https://twitter.com/settings/account#personalize_timeline and uncheck "Show best tweets first". It will still inject a block called "In case you missed it" from time to time but you can dismiss it.
Mastodon... it's a platform, not a single entity that I can use.
That might be the tradeoff that has to be made. Pick a Mastodon instance whose administration you trust not to screw you over. If they do you can move to another. If you go with a single entity, you're going to be stuck when they mess with you. Like with Twitter.
Well, it goes you a chronological ish timeline. It still insists on trying to insert blocks of "content you might have missed", and it still has crappy conversation views that can't be turned off.
A strictly chronological timeline is only available via third party apps.
Plus "Promoted Tweets". Those are also out of order, and increasingly just as irritating as the "content you might have missed" blocks.
Also, now the "content you might have missed" stuff is showing up in Notifications and whoever thought that was a good idea?
That only seems plausible if most of the people I want follow are already there.
Twitter had that chicken-and-egg problem too, but do you know how we solved it? We all joined Twitter and then told people we wanted to follow that we were there.
With the difference, that back then Twitter didn't exist. Only in form of IRC or such and Twitter could distinguish from that. A new service has to be unique in a new way.
Like not screwing its users over?
>creating a Mastodon server
This is really what Mastodon needs, a solid sponsor. I want to use it, but the loose federation model without core instances that have a critical mass of users makes it hard to commit to. In Japan, when Pixiv sponsored Pawoo, the Mastodon usage shot up, I think that's still one of the biggest instances worldwide right now.
Also, Mastodon needs a good Mac client. I would pay for both that and to support a general Mastodon instance backed by someone with name recognition.
What if Twitterific and Tweetbot just default to posting to both platforms, users could turn it off, to kind of dark pattern posts onto Mastadon or another alternative.
This might not be an area people want to discuss, but at this price does it make a grey market API competitor viable? Meaning first party API access is seen as preferable because HTML scraping is unreliable and therefore necessitates a full time staff of individuals to respond to site changes.
But if the price is high enough, couldn't a third party operating out of another part of the world undercut them using scraping, a dedicated team, and a clean API? Twitter may have technical remedies but it is unlikely they'd have legal ones outside of the US and a few Western countries.
Official APIs Vs. "pirate" APIs haven't really been a major issue until now because companies simply weren't charging enough to make the alternative cost effective. Is that going to change with pricing strategies like this one?
These sorts of 'grey market' APIs already have a lot of infrastructure support too. Paid APIs haven't been the push, but people have wanted programmatic access to other peoples' site data for a while now. The real driver has just been "no API", not "pricey API".
Is there an industry term for non-official APIs? I often find myself wondering if a third-party service already exists around some aspect of a platform's "crown jewels", but don't really know how to search for them.
Not that I know of, I just know that there's an ecosystem of bespoke site wrappers that people will pay for, and some folks who are quite good at it. Kind of grey hat stuff and it's been a few years since I was in contact with the scene, if you could call it that.
Downloading a webpage and returning a snippet of it via API isn't greyhat work, just like building a search engine isn't greyhat work. Processing public websites is white hat work, and very common too! If a business wants to restrict access, they can take down their website.
Any company that has anything valuable that anyone else would want is going to put TOS there prohibiting scraping, making it grey hat and legally risky.
You don’t even have to pay. Just disassemble one of the official apps, snatch the private keys and use them to connect to the API. Enjoy having access to everything with (almost) no rate limiting.
That can be mitigated easily by rotating keys in every app update.
You could continuously push out keys at when they do…
It looks like this is going to hit clients like Twitterific/Tweetbot pretty hard https://twitter.com/BigZaphod/status/996784838349910016
It's sad to see how Twitter is treating 3rd party devs, and "engaged" users — the kind of users willing to buy a 3rd party app to use their services. Furthermore, these engaged users bring their value to the platform by bringing in eyeballs from less engaged audiences rather than clicking on ads themselves.
For Twitter third party clients are basically competition so it's not surprising that they make API access as limited as possible. Furthermore because of network effects they are not afraid that a significant number of people will leave the platform.
Well, they could try to compete and make decent apps first.
For some reason they're treating them like competition.
Hopefully they'll actually become competition and push users (the engaged ones who fed twitter to its current state and are dissatisfied now) to use Mastodon.
As long as they're showing the ad-tweets, they're not competition.
I have no baseline to go by, but my gut feel is that the price is quite high. If it isn't, that means my data is worth $12 a month. Think of how many services are scraping/selling/aggregating/whatever-ing your data. Each one aught to be paying you the equivalent of $12 a month, either in straight cash for something you pay for, or in services.
I know I'm not getting my value.
I think the context on this is http://apps-of-a-feather.com/ on the recent changes to their api for app developers.
Years ago they raised the minimum API entry pricing. I used to pay for Topsy and they gave aggregate data that was quite insightful and reasonably priced.
Soon after Apple's acquisition they (1) jacked up the price, and then (2) shut it down soon after.
The reality is that Twitter doesn't have the resources allocated to manage tiny customers and doesn't desire to. That decision was made years ago. They can give an employee 10 customers to manage each of whom spend $36,000 a year and likely will increase and the math adds up.
Is anybody really shocked by this pricing?
Yes, because the platform is a worldwide social network, not a B2B app. Also what is there to manage? There's no need for the vast majority of app developers to talk to anyone to use an API.
OP has a legacy mindset, essentially the only time I talk to a company about their API is when it returns bad data, or is non-functional. Outside of that, I might interact with an account manager at onboarding (due to some companies forcing you to deal with one), and if the API is heavily used, I may reach out and ask 'em if they can go cheaper, while shopping around with other companies offering APIs for said purpose.
I have a Twitter account that retweets a specific # from followers (who submit crowdsourced train updates). Would these count as 1 subscription, or how does the subscription counting work?
On the same account, I have some very basic chatbot thingy where you can ask "where is train 1234?" and get a response. It works by listening to the User Stream, and using regex to check if the "bot" is being asked a question.
> When we announced our plans to retire and replace Site Streams, User Streams, and our REST Direct Message endpoints, we did so as part of our efforts to improve the developer experience and provide a sustainable way to help businesses engage with those looking to connect with them.
Streams were pretty suitable for my use-case, so I don't believe Twitter on this sentence.
Their changes seem like they just want a way to get non-firehose users to pay for accessing data streams, and might alienate some developers. Someone else mentioned the short timeline over which to migrate (for devs that only get access from today).
Dear Twitter, you've had a bad run with alienating developers in the past; please decide what you want to do. If we build our communities around your platform, be more clear about what we have to now pay for.
All I need to know is which 3rd party provider will give me a linear timeline, even if I have to pay a little. Twitter refuses to stop "curating" my feed on their website, so clearly I need to rely on a 3rd party. Facebook has long since become utterly useless due to feed curation.
The founders of thirt party apps already think it won’t be financially viable for them and their users.
The thing is "a little" might be on the order of 10 bucks a month.
I think its important to qualify “users” in this case. Technically, its 250 subscriptions, whether or not theres a one to one mapping between subscription and user depends on the use case, although for a an app where each user requires their own unique stream, this is likely the case.
Seems to be how Twitter is accommodating for twitter clients not showing ads. That's like $11 a month for the 250 users but I imagine that the enterprise could get as low as $5/user/month.
It's almost like the ad-free experience people want on social networks would be possible with this. I'm hoping they work with Tweetbot and this isn't a passive aggressive way to shut it down.
This I think is their enterprise data service as powered by a company they acquired awhile back -- I mean, for me if I were running a business an enterprise customer is at least this amount of money per month or otherwise why bother with them at all. Not sure what the big deal is.
I’m just going to sit and wait until Twitter falls into the irrelevancy black hole.
They very nearly did until Trump came into power. Remember all the articles talking about the troubles twitter was having and everyone predicting the end. And then the campaign happened. Now everyone has twitter just to see what he'll say next.
So, this is what they’re going to ask Tweetbot and the rest of the alternative Twitter clients to switch to?
Here's Sean Heber's response to this: https://twitter.com/BigZaphod/status/996794900208287744
It seems like they're waiting for enterprise pricing: https://techcrunch.com/2018/05/15/tweetbot-3-arrives-with-a-...
I'd pay $10/month for Tweetbot. It is an ad-free twitter experience after all.
Well Tweetbot will be paying Twitter $11/mo. For them to also make some money, they have to charge you closer to $20/mo.
Tweetbot will not pay that, they'll poll instead of stream and have a worse product because of it.
Tweetbot with poll still beats Twitter.com with streaming.
I think polling instead of real-time. The timeline has decent rate limits, but I can't understand if the Direct message is rate-limited globally for the app or not?
That is absolute insanity...
At the very minimum.
Anything for monitoring tweets with hashtags?