I thought I read that the Google Cloud thing happened at least one week before the issues, and the bills were quickly paid. Additionally, Twitter runs its own servers and Google Cloud is only used for analytics and side stuff. The relationship between the rate limit and all that seems like a hoax.
Yeah, my impression was that this started with Twitter's decision to stop showing tweets to people who aren't logged in (from June 30th), which I imagine was in response to people circumventing the high API costs by just scraping from the website. And I'm speculating that this had unexpected effects because none of the software was written with this in mind, which made Twitter really slow. Or maybe the scrapers switched to scraping via newly created accounts and this somehow overloaded Twitter (?) though I don't really see why this would require more bandwidth than scraping without an account.
I still am baffled by people who are baffled at the refusal to pay.
This article itself mentions that the platform's management is erratic and unpredictable, that the rules change on a whim, and that availability is hit-or-miss at best. In the light of that, why would *anyone* pay? This observation applies to advertisers as much as it applies to users.
My model is that the various draconian measures are _exactly_ the kinds of 'bandaids' that are applied in an 'emergency', i.e. before a better longer-term solution can be designed, implemented, tested, and deployed. And Elon in particular is a fucking top-tier (if not literally #1) example of 'failing fast' and being willing to break things as an explicit and deliberate tradeoff for higher 'velocity'/'tempo'.
I also expect Paul Graham is being a little glib, or maybe naïve (or out of touch with concrete software maintenance/admin work), about, e.g. the ease by which scrapers can be reliably distinguished from 'human users'. One reason is that the existing Twitter clients, e.g. the website itself, the official Twitter apps, might be _fantastically_ inefficient and basically 'bad API users'. (Reddit's own apps are exactly this terrible, which was extra funny to learn during their own recent drama.)
I've been a much more active Twitter user over the past few weeks. (I'm a Reddit refugee.) It has been almost entirely fine – for me. I saw one seemingly spurious 'you have liked too many tweets' notice. The constant complaining is grating, tho it doesn't even seem _that_ much more common above baseline (outside of the specific people/accounts I'm following).
I was, and am, very Sad that 'just don't pay your bills with your most important vendors' is such a common 'business tactic'. I'm not sure I even have a good (single, unique) model of how to feel and think about it! I'm sure there are 'legitimate' scenarios in which it both makes 'business sense' and is at least 'morally defensible', but the boundaries seem VERY fuzzy. I guess it's not _really_ that much different than 'aggressive negotiation of billing terms', in some/many sense(s), and my own personal (professional) willingness to 'fire clients' kinda sorta makes me want to 'round it off' to 'risk management' or something similar! I still have some visceral distaste of the (seemingly too-common) practice.
Intriguing predicament! I appreciate your swift adaptation, John. This just goes to show how swiftly we must evolve with tech disruptions, including AI. I recently stumbled across a unique example - a tool that leverages AI for enhancing self-portraits. It's quite the marvel and leads me to wonder how such technology could benefit various digital platforms, especially during chaotic times like these. An interesting confluence of tech and user experience, wouldn't you agree? Have a peek here - http://dating.tiktak-studio.com. The future certainly is a curious place.
I'm pretty sure this is spam, but it's the first one like it and so well-done that I am leaving it here. Interesting unforced error to say John here. Future attempts on this level will get deleted silently, though.
However, if anyone wants to pitch me that I should check out their (or someone else's) AI service cause it's cool, you are free to do that - just be explicit that this is what you're doing.
As a card-carrying "John" I can say it sure looks like spam to me.
So that my comment isn't totally useless, I deleted my twitter account a while back as it was getting just a bit too hot for me emotionally on there (and adversely affecting my health). I didn't tweet much but did respond to some of your polls. After deleting my account, I still "followed" some profiles via unauthed browser tabs, now it looks like I can't do that anymore, so its the end for twitter and me.
I wish there were some way to pay for Twitter Blue without giving Twitter my real name and phone number. Twitter's track record in information security leaves much to be desired.
Tweetdeck changes have been pretty disastrous for me. I might be willing to shell out for blue purely for Tweetdeck if it had all of the features old Tweetdeck had with or without improvements, but I had mine configured in particular ways that are simply not possible now. Had mine set up with a column for likes and follows from people I’m following and I’d always been able to find great stuff (and new people to follow) by having that in its own stream separate from the generic default timeline of just posts and retweets. As is, it feels like they just took the generic existing desktop app, made it so you could cram them into multiple columns, and called it good. Strong vibes of the sentiment that they had to gut the previous Tweetdeck because it was clearly breaking whatever they tweaked and didn’t actually have a replacement for it in development.
For now, I’ve settled for using a chromium add-on that reverts Twitter to the way old circa 2015 version. It’s not exactly perfect either but better than current and for everything that breaks it has a “see this page in current Twitter” button that works around stuff.
Since you didn’t mention in “The Plan” I thought I’d point out that posting Twitter screenshots the way you have been is a pretty acceptable alternative to linking tweets IMO, granted less tenable for longer threads or when a back and forth doesn’t fit cleanly into one thread. As it was I would occasionally click those links only to find out it was the same as the content posted in the accompanying screenshot anyways.
I may definitely be post-hoc justifying out of pure stubbornness here but I don’t really feel like Twitter deserves my money for the quality of product we get, arguably even inasmuch as you can reduce it to something like 10 cents per hour. The value of Twitter is almost entirely product of it being society’s chosen commons (in a Georgian-esque, the-citizens-are-creating-the-real-estate-value-not-the-landlords way), and while I would agree they should perhaps be rewarded for having the metaphorical internet real estate that we use to conduct our discourse, their reward is in easier access to advertisers not direct access to my wallet. The level of experience I get as an individual (especially one with virtually no engagement even before blue changed things) is not subject to the level of competitive constraints that give me an experience I think deserves to be paid for. The ideal situation is that I would have a number of interfaces through which to access Twitter’s content and could chose to pay for whichever I thought was the best, but this is the opposite of the present situation where Musk seems to very much want us to experience Twitter an extremely constrained single way.
I consider Twitter a chat, akin to IRC, where people type short messages that are mostly noise, but since they're short and take only one line, I can quickly scroll through them and filter out the signal.
I've tried multiple different solutions to reading Twitter, and the only one I could keep up with was a Slack bridge, that changes the Twitter UI to exactly that, a chat with one tweet per line (with sometimes image and video embeds in between).
When they took the API from me, there was no simple changes I could make to restore my workflow. I could've resorted to scrapers, but I'm glad I didn't. So I've had to stop using Twitter as a primary source and I only get updates indirectly from places such as this blog post.
Mostly due to this I'll gladly take a transition to Bluesky or Threads, as they both (currently) promise interoperability.
Although since it seems that I am alone with this "Twitter is a chat" opinion, I'll still have to rebuild the workflow myself if there is a transition.
So I happened to notice this morning that individual tweets are once again visible without being logged in but it appears to only shows the directly linked tweet and no replies. Clicking anywhere else prompts a login except the person's profile which loads but none of their tweets load. Probably still in flux currently.
I thought I read that the Google Cloud thing happened at least one week before the issues, and the bills were quickly paid. Additionally, Twitter runs its own servers and Google Cloud is only used for analytics and side stuff. The relationship between the rate limit and all that seems like a hoax.
Yeah, my impression was that this started with Twitter's decision to stop showing tweets to people who aren't logged in (from June 30th), which I imagine was in response to people circumventing the high API costs by just scraping from the website. And I'm speculating that this had unexpected effects because none of the software was written with this in mind, which made Twitter really slow. Or maybe the scrapers switched to scraping via newly created accounts and this somehow overloaded Twitter (?) though I don't really see why this would require more bandwidth than scraping without an account.
I agree that I came out a little too strong to blame loud in particular but that this is about compute costs seems super clear, one way or another.
Glad I'm still using RSS feeds and a good feed reader. Maybe everyone abandoned that technology a little early.
Rip Google reader. What feed app do you use?
I pay for NewsBlur. It's something like $12 a year.
I use Feedly, for everything that plays well with RSS. Not a match for Twitter of course.
Twitter, or similar, are VERY different than using a feed reader IME. (I LOVED Google Reader – had the best of 'both worlds' to a large degree.)
Musk has altered the deal, prey
*Pray
(Should I not do this? But if you edit Twitter threads you probably care?)
Notes on substack isn't even mentioned. I have found it extremely slow.
I still am baffled by people who are baffled at the refusal to pay.
This article itself mentions that the platform's management is erratic and unpredictable, that the rules change on a whim, and that availability is hit-or-miss at best. In the light of that, why would *anyone* pay? This observation applies to advertisers as much as it applies to users.
My model is that the various draconian measures are _exactly_ the kinds of 'bandaids' that are applied in an 'emergency', i.e. before a better longer-term solution can be designed, implemented, tested, and deployed. And Elon in particular is a fucking top-tier (if not literally #1) example of 'failing fast' and being willing to break things as an explicit and deliberate tradeoff for higher 'velocity'/'tempo'.
I also expect Paul Graham is being a little glib, or maybe naïve (or out of touch with concrete software maintenance/admin work), about, e.g. the ease by which scrapers can be reliably distinguished from 'human users'. One reason is that the existing Twitter clients, e.g. the website itself, the official Twitter apps, might be _fantastically_ inefficient and basically 'bad API users'. (Reddit's own apps are exactly this terrible, which was extra funny to learn during their own recent drama.)
I've been a much more active Twitter user over the past few weeks. (I'm a Reddit refugee.) It has been almost entirely fine – for me. I saw one seemingly spurious 'you have liked too many tweets' notice. The constant complaining is grating, tho it doesn't even seem _that_ much more common above baseline (outside of the specific people/accounts I'm following).
I was, and am, very Sad that 'just don't pay your bills with your most important vendors' is such a common 'business tactic'. I'm not sure I even have a good (single, unique) model of how to feel and think about it! I'm sure there are 'legitimate' scenarios in which it both makes 'business sense' and is at least 'morally defensible', but the boundaries seem VERY fuzzy. I guess it's not _really_ that much different than 'aggressive negotiation of billing terms', in some/many sense(s), and my own personal (professional) willingness to 'fire clients' kinda sorta makes me want to 'round it off' to 'risk management' or something similar! I still have some visceral distaste of the (seemingly too-common) practice.
Intriguing predicament! I appreciate your swift adaptation, John. This just goes to show how swiftly we must evolve with tech disruptions, including AI. I recently stumbled across a unique example - a tool that leverages AI for enhancing self-portraits. It's quite the marvel and leads me to wonder how such technology could benefit various digital platforms, especially during chaotic times like these. An interesting confluence of tech and user experience, wouldn't you agree? Have a peek here - http://dating.tiktak-studio.com. The future certainly is a curious place.
I'm pretty sure this is spam, but it's the first one like it and so well-done that I am leaving it here. Interesting unforced error to say John here. Future attempts on this level will get deleted silently, though.
However, if anyone wants to pitch me that I should check out their (or someone else's) AI service cause it's cool, you are free to do that - just be explicit that this is what you're doing.
As a card-carrying "John" I can say it sure looks like spam to me.
So that my comment isn't totally useless, I deleted my twitter account a while back as it was getting just a bit too hot for me emotionally on there (and adversely affecting my health). I didn't tweet much but did respond to some of your polls. After deleting my account, I still "followed" some profiles via unauthed browser tabs, now it looks like I can't do that anymore, so its the end for twitter and me.
I wish there were some way to pay for Twitter Blue without giving Twitter my real name and phone number. Twitter's track record in information security leaves much to be desired.
Tweetdeck changes have been pretty disastrous for me. I might be willing to shell out for blue purely for Tweetdeck if it had all of the features old Tweetdeck had with or without improvements, but I had mine configured in particular ways that are simply not possible now. Had mine set up with a column for likes and follows from people I’m following and I’d always been able to find great stuff (and new people to follow) by having that in its own stream separate from the generic default timeline of just posts and retweets. As is, it feels like they just took the generic existing desktop app, made it so you could cram them into multiple columns, and called it good. Strong vibes of the sentiment that they had to gut the previous Tweetdeck because it was clearly breaking whatever they tweaked and didn’t actually have a replacement for it in development.
For now, I’ve settled for using a chromium add-on that reverts Twitter to the way old circa 2015 version. It’s not exactly perfect either but better than current and for everything that breaks it has a “see this page in current Twitter” button that works around stuff.
Since you didn’t mention in “The Plan” I thought I’d point out that posting Twitter screenshots the way you have been is a pretty acceptable alternative to linking tweets IMO, granted less tenable for longer threads or when a back and forth doesn’t fit cleanly into one thread. As it was I would occasionally click those links only to find out it was the same as the content posted in the accompanying screenshot anyways.
I may definitely be post-hoc justifying out of pure stubbornness here but I don’t really feel like Twitter deserves my money for the quality of product we get, arguably even inasmuch as you can reduce it to something like 10 cents per hour. The value of Twitter is almost entirely product of it being society’s chosen commons (in a Georgian-esque, the-citizens-are-creating-the-real-estate-value-not-the-landlords way), and while I would agree they should perhaps be rewarded for having the metaphorical internet real estate that we use to conduct our discourse, their reward is in easier access to advertisers not direct access to my wallet. The level of experience I get as an individual (especially one with virtually no engagement even before blue changed things) is not subject to the level of competitive constraints that give me an experience I think deserves to be paid for. The ideal situation is that I would have a number of interfaces through which to access Twitter’s content and could chose to pay for whichever I thought was the best, but this is the opposite of the present situation where Musk seems to very much want us to experience Twitter an extremely constrained single way.
Can't hurt to make a Bluesky account just in case twitter continues going down the tubes in order to have somewhere not run by meta to go.
I consider Twitter a chat, akin to IRC, where people type short messages that are mostly noise, but since they're short and take only one line, I can quickly scroll through them and filter out the signal.
I've tried multiple different solutions to reading Twitter, and the only one I could keep up with was a Slack bridge, that changes the Twitter UI to exactly that, a chat with one tweet per line (with sometimes image and video embeds in between).
When they took the API from me, there was no simple changes I could make to restore my workflow. I could've resorted to scrapers, but I'm glad I didn't. So I've had to stop using Twitter as a primary source and I only get updates indirectly from places such as this blog post.
Mostly due to this I'll gladly take a transition to Bluesky or Threads, as they both (currently) promise interoperability.
Although since it seems that I am alone with this "Twitter is a chat" opinion, I'll still have to rebuild the workflow myself if there is a transition.
So I happened to notice this morning that individual tweets are once again visible without being logged in but it appears to only shows the directly linked tweet and no replies. Clicking anywhere else prompts a login except the person's profile which loads but none of their tweets load. Probably still in flux currently.