The force, facing "unprecedented budget cuts," intends "to allow the public to see what officers at one of the largest UK forces face on a daily basis," with the tweets, but apparently the force faces so much that it goes well beyond what Twitter allows.
The experiment started at 5 a.m. GMT and it looks like they quickly began running into rate limit problems. Twitter tries to stop users from overloading the service by limiting the rate at which they tweet. The @gmpolice account has attempted to avoid this issue by using three accounts, @gmp24_1, @gmp24_2 and @gmp24_3.
Two police staff from the force's public relations department are responsible for the tweets, which are cataloging in near real-time all 999 emergency-line and switchboard calls. So far, the number of calls have been overwhelming it seems, and the staff have to continually switch from one account to another to "avoid Twitjail", or having the account temporarily shut down due to these rate limits.
Rate Limits and Innovation
Twitter's rate limit limits the number of times a third-party application can make requests via the API in a given time period. Tweets, however, are not specifically rate limited, but limited nonetheless. As ReadWriteWeb's own Web guru, Jared Smith, puts it, "It's likely they are hitting the semi-hourly tweet limit, which is fuzzy and vaguely documented."
When rumor hit last year that Twitter would be increasing the rate limit ten-fold, Marshall Kirkpatrick argued that "current API limits are a constraint on how much analysis you can perform, bake-down and present to your users." With all of the emphasis we put on the real-time Web, it feels like something is wrong if an experiment as simple sounding as this one needs multiple work-arounds and keeps running into walls. This is it, right here - this is the real-time Web... a half hour later and somewhere else.
Bask in the Hashtag
Fortunately, the entire thing is strung together though the "#gmp24" hashtag, so a continuous stream is available on the user-side. The hashtag not only makes it possible to pick out the innumerable gems, but also gather all related tweets into one stream for later data-mining.