Is Twitter Winning the War on Spam? Our Stats Do Not Support this Assertion

Frederic Lardinois’ article on March 23, 2010 refers to Twitter’s assertion that spam is not an issue. According to the latest data from Twitter, the percentage of spammy tweets per day is now down under 1%.

Here’s the issue. Our stats don’t support this assertion.

In fact, our numbers suggest that spammers are alive and well on Twitter. How do we arrive at this conclusion? In December of 2009 we launched TwitSweeper. Its only function is to detect undesirable followers (spam and spammers), identify them, and list them so TwitSweeper users can choose to remove and block them. It does this very well.

Here’s what we know.

To date TwitSweeper has checked more than 1.7 million followers. It has detected and flagged more than 83,000 followers as being spam or spammers. This suggests that almost 5% of all the followers we’ve checked are spammy.

The next question is: What is Spam?

The universal answer seems to be: “Everything I don’t want to receive from someone I don’t know.”

The problem with this definition is that it can include everything. Promoting body enhancement products is a nuisance to most, but a blessing to some. One man’s garbage is another man’s gold.

Is Twitter doing a good job going after spammers? I’m sure they are making best efforts. Are they winning the battle? It’s too early to say. Do they have spam down to 1%? Our stats certainly don’t confirm that. If our stats indicate that 5% of followers are spammers, you’ve got to know the actual number is even higher.

Conclusion: As TwitSweeper’s Spam Detection Methodology continues to improve, we see a corresponding rise in spammy followers that are found. This suggests there are more spammers out there. Whereas, Twitter’s reported 1% spam figure is merely the portion of spam that they happen to have found so far.

In the end, everyone’s objective should be to keep the Twitter ecosystem clean of spam. But we aren’t there yet.

3 Responses to "Is Twitter Winning the War on Spam? Our Stats Do Not Support this Assertion"

  1. Dave
    March 25, 2010 at 1:01 pm #

    Two different stats: Twitter is counting messages as a percent of total, you’re counting users as a percent of your subset of total.

    I’m not saying one is more accurate than the other, just that they’re different.

  2. Doug Braun
    March 25, 2010 at 4:01 pm #

    Agreed. Both are valid measures of spam “stuff” as a percentage of its respective total.

    The 83,000 (and growing) spammy followers that has found is 5% of the 1.7 million followers that TwitSweeper has checked thus far.

    I guess you could ask yourself if (motivated/automated) spammers would tweet more frequently or less frequently than the average Twitter user, if you wanted to look at or convert the “percentage of followers” measure into a percentage of tweets.

    Yes, two different ways of measuring spamminess as a percentage of a total. We just don’t think that spam on Twitter, in any measure, is yet under control nor that it is a low as 1%. The TwitSweeper numbers don’t seem to support such a conclusion.

  3. Savannah
    March 25, 2010 at 10:31 pm #

    I don’t know how much there is, but I know the spam there is disheartening. I use twitter for my non-profit work and when I recently ran a search on “adult literacy” the first 3 pictures to pop up were porn bots with severely inappropriate pictures. What’s a NFP professional to do?

Post a Comment

Your email is never shared. Required fields are marked *