Those of you who use Twitter know that the folks at Twitter haven’t been able to keep the service running properly.  You may also know that in an effort to combat the enormous amounts of negative PR that they were receiving as a result the downtime, they started the Twitter Status Blog, which keeps you updated on every bump and hiccup the system has.

Yesterday, Twitter announced that during the Stevenote, Steve Jobs’ keynote address at the Apple Worldwide Developers Conference (WWDC), that some features of Twitter would be disabled and that API requests would be limited to 10 per hour.  The Twitter team did this to keep Twitter running during a period when they expected 10x their normal traffic levels.  Apparently, the steps they took worked.  While this served to keep Twitter running, it essentially brought the system down for those of us who rely on the API to use the service.  Anyone who uses Flock, twhirl, AlertThingy, or any of the other myriad of Twitter apps was only able to receive updates every 6 minutes.  considering the conversational nature of the tool and the frequency with which I receive updates (probably 5-10/minute during the WWDC), this made Twitter virtually useless to me.

As you can probably tell, I was not (and am still not) very pleased with the way that Twitter decided to cope with the stress on the system.  I just shut Twitter off yesterday.  Later, the Status Blog mentioned that they had increased the request limit to 20 requests per hour, but this still wasn’t enough.  Twitter requires separate requests for messages, @ replies, and direct messages, meaning that I could still only get updates about once every 4 minutes, and I could only request @ replies and direct messages once per hour.  Remember that the original API limit was 70 requests per hour; then they dropped it to 30.  This change, done without any announcement until well after it was implemented, caused trouble for many users.

That change was made before Twitter started their crusade to be open about the problems they are facing, but Twitter recently made another change without announcing it.  At some point yesterday, they set the API request limit back to 50 per hour.  This is a very welcome change.  While I hope that the final solution to Twitter’s scaling problems isn’t to limit functionality during big events, I am pleased that they have increased the request limit and hope that they restore it to its original level soon.

Note: This means that you can reconfigure your client to request your tweets more often.  Instructions for twhirl are here.

---

Since most of you probably don’t care why your twitter client (i.e. twhirl, AlertThingy, etc.) isn’t working and just want to know how to fix it, I’ll go through the fix first.

If you’re getting a “limit exceeded” message or if you simply stop receiving Tweets, you need to tell your client to request your Tweets less often.

Changing this setting is pretty easy to do on twhirl. Just click on the wrench icon at the top of the window (directly to the left of the minimize icon), and go to the “Network” tab. Set the “Request usage” option to 30 requests per hour or less. You can then fine tune how often you get Tweets, DMs, and @ replies. I tend to want my Tweets more frequently (every 3 minutes) and replies and DMs less often (every 20 minutes), but you can obviously fine tune this depending on how you use Twitter.

I’m not sure how to do this on other clients (AlertThingy still really baffles me), but you can probably find the setting if you hunt around a bit.

—

For those of you who do care about the cause behind this problem:

Twitter recently changed their API to lessen the stress on their system.  If you use Twitter with any regularity, you know they’ve been having a lot of downtime recently.  Basically, the non-IM clients were previously allowed to access your Twitter account 70 times per hour, but the folks at Twitter limited these requests to 30 per hour. Consequently, you have to tell your client to request Tweets less often. Otherwise, about 30 minutes after you first open your client it will stop working for about 30 minutes, and this 30 minutes on/30 minutes off problem will continue until Twitter moves the limit back to 70 requests per hour.

Moreover, my (not personally verified) understanding is that the IM clients (GTalk, AIM, etc.) aren’t working at all right now, as Twitter has disabled the Jabber/XMPP protocol.

Hope this helps!

---