Twitter has published figures that appear to show a dramatic fall in spam on the service.
In a blog statement, the company reckons that the number of ‘spammy tweets' posted per day is now around 1 percent and falling, a sharp drop compared to the high point last August of between 8 and 11 percent.
"We're constantly battling against spam to improve the Twitter experience and we're happy to report that it's working," says the company, which now employs a special Trust and Safety team just to battle the problem.
The exact definition of spam on Twitter involves detecting more complex patterns of behaviour than it would on email. The most obvious problem is the setting up of bogus accounts which then generate tweets or direct messages that lead to malicious websites, but also the repeated following and ‘unfollowing' of users in order to attract attention, Twitter says.
The service is not immune to direct hacking either, as the arrest in recent days of a man accused of hacking President Obama's account demonstrates.
The company's main weapon in the battle is probably the growing sophistication of its core users who are now more willing to take the time to report spam. Earlier this month the company announced that it was now filtering shortened links used in direct messages and tweets.
According to Barracuda Networks, the spam peak last August on Twitter was driven by the service reaching a sudden tipping point after key celebrities joined. This drove criminals hoping to piggyback on interest in these individuals to push more links.
Barracuda's own estimates suggest that Twitter last enjoyed a spam rate as low as 1 percent in 2006, perhaps only weeks after it was founded. Given that the service now has tens of millions of users, the absolute volume of spam is now massive even if the proportion is small.