Tuesdays in December worst for security threats

Simon · Jan 10, 2011, 22:26:52

Tuesday is the worst day for security threats around the world, according to a report from SonicWall.

The security firm said Monday came close behind Tuesday as the "most active day for threat-related traffic" for much of Europe, the US, China and India.

In the UK specifically, Friday was the worst day, at least for the week in November that SonicWall released data for.

Back to school time and the Christmas holidays were the worst hit by security threats. Trojans peak in September and December, while worms wreak the most havoc in December, and adware is at its worst during September, October and December, the report said.

Read more: http://www.pcpro.co.uk/news/security/364234/tuesdays-in-december-worst-for-security-threats

Technical Ben · Jan 11, 2011, 08:20:53

Is this in correlation with windows update? I have a feeling they release updates on Tuesdays. So it naturally follows that the existing bugs give one last push, and new ones emerge to get through new holes (or the ones missed).

Rik · Jan 11, 2011, 10:45:50

Seems likely, Ben. I wonder why we are Fridays though?

Simon · Jan 11, 2011, 10:51:47

Because some support services close for the weekend?

Rik · Jan 11, 2011, 10:52:15

No names, no pack drills?

Simon · Jan 11, 2011, 10:58:48

But some wouldn't break into their holiday to help.

Rik · Jan 11, 2011, 10:59:40

armadillo · Jan 11, 2011, 23:28:05

Quote from: cited linkIn the UK specifically, Friday was the worst day, at least for the week in November that SonicWall released data for.

If the rest of the report is as statistically strong as that, it would make entertaining reading.

Take one week. Arrange the figures in order of size. One day is on top. The day has a name. The name was Friday. So what?

The statement about Fridays makes me doubt whether they did any statistical analysis at all.

If events occur randomly spread across days (or months), then the probability of all day names having exactly the same number of events is vanishingly small. It follows that it will be possible to arrange them in order of size. It takes a fairly large sample and some rigorous statistics to establish whether there is any correlation between day name and number of events. They may have done all that. But somehow, I doubt it.

Technical Ben · Jan 12, 2011, 09:53:37

WHAT!?
They only tested one week.

That is a 100% epic fail for "week on week" or comparison. To mention per day, with a sample rate of 1 (only checking 1 Friday) gives you a possible error margine of 100%.