Why the discrepancy? First of all, not all researchers study botnets from the same perspective. Some focus more on command and control, others at the malware itself, and others on the potential zombies (plus each has its own tools). Sure, the numbers associated with this infection are mind-boggling and scary, but even with the disparity between 500,000 and 8 million, all botnet hunters are watching it very, very closely.
Botnet counts and classification disagreements are nothing new in the research community. Remember last year's Bobax/Kraken debate? One group of researchers called it a new botnet, while another said it was an existing one. Now we just call it Bobax/Kraken to split the difference.
What the botnet researchers do agree on with Confickr/Downadup is that it has the makings of a botnet. The infected machines are communicating with an average of 250 different domains each day, attempting to download more malcode. But those domains mostly have been unregistered, with no set group to indicate that a botnet command and control infrastructure is in place.
Just what the bad guys have planned for this astronomical number of machines, assuming the 8.9 million number is the most accurate, is worrisome. So far, it has been all about pushing rogue antivirus software, but just think of the damage a botnet of this size could do. It wasn't long ago that researchers had written off former super-sized botnets like Storm as too high-profile and unwieldy to survive. Now Storm is back, although far smaller than its former size, and a possible new botnet is brewing.
-- Kelly Jackson Higgins, Senior Editor, Dark Reading