Wednesday 2 February 2005

Spam volumes redux


Several anti-spam companies talk about spam volumes in terms of a percentage of all inbound mail. Outsourced anti-spam services such as BlackSpider and Postini are currently quoting spam volumes in the 70%-85% range, having steadily grown over the last two+ years.

That's nice, but it's actually hard to grasp what that means in absolute terms. All you maths graduates will immediately realise that spam percentage can never grow to 100% (unless people stop sending legitimate email of course).

For the rest of, who struggle to visualize asymptotes and such, here are a couple of graphs to illustrate this...

The first graph shows the usual percentage figures that we're used to seeing from anti-spam companies. It plots the growth of spam since the start of 2003. I've smoothed the peaks and troughs for clarity. As some vendors have recently implied, it looks like the volume of spam is starting to level off.

Now look at the second graph. It shows the same data, but expressed as the number of spam messages. Now we see that, far from slowing down, the volume of spam continues to grow at a fairly consistent rate. In fact, I calculated these numbers assuming that the level of legitimate email stayed static, which would probably understate the spam growth. (Note there's deliberately no Y-axis scale on this graph: the shape is what's important, not the scale.)

So tell me again... is spam levelling off?

Trivia from the MIT Spam Conference


MIT held its 2005 Spam Conference last week. Some interesting talks, with new ideas. Some mad-sounding, academic ideas that would never work in the real world. Unfortunately some boring, unoriginal, or marketing-driven ones too.

Some interesting trivia:

  • There are 600,426,974,379,824,381,952 ways to mis-spell \/¡äğŗă
  • About 5% of web traffic appears to be spammers harvesting email addresses
  • On average, spammers take about 10 days to send spam to a newly-harvested address
  • At one point, an audience member asked, "If the opposite of Spam is Ham, what's the opposite of Phish?" Quick as a flash, the answer came back... "Phowl."