16 comments

[ 3.1 ms ] story [ 47.7 ms ] thread
From Wikipedia:

> On April 21, 2010, beginning at approximately 14:00 UTC, millions of computers worldwide running Windows XP Service Pack 3 were affected by an erroneous virus definition file update by McAfee, resulting in the removal of a Windows system file (svchost.exe) on those machines, causing machines to lose network access and, in some cases, enter a reboot loop. McAfee rectified this by removing and replacing the faulty DAT file, version 5958, with an emergency DAT file, version 5959 and has posted a fix for the affected machines in their consumer knowledge base.

BBC article from the time: http://news.bbc.co.uk/1/hi/technology/8636985.stm

(comment deleted)
Quite a few similar submissions in the last 24 hours gained enough upvotes to show on the front page, seems like someone keeps flagging them. I'm not sure why though, I think the discussion about the similarities and potential reasons behind these events could prove interesting.

This post is not even on the first four pages despite ~20 upvotes within an hour, from experience this takes quite a few flags

Yes would be interesting to know what policy implementations or directions could have been common factors between the two companies which led to these incidents.
Yes, I'd be interested to hear that. From initial reports, it sounds like under-investment in QA because of over-confidence in automated QC.

It's an example of a "black swan" or Bertrand Russell's chicken - the same process has worked many times, leading people to make the false conclusion that the risk has become neglible. There's a successful trial period in which the beancounters reduce headcount with no negative consequences. So the trial becomes permanent, people become less careful. And then boom.

They miss that a 1 in 10000 occurence is going to happen eventually, and that "unnecessary expense" which was previously there to mitigate it has now been removed.

https://news.ycombinator.com/item?id=41007084

The similarities are basically coincidence - the rather confected connection doesn't tell you much about either event and the most recent one is in active discussion on the front page already.

The significance may be confected, but the connection certainly is not, being a matter of public record.
Sure, we can slice that way if you prefer it but it doesn't make it a more interesting or non-repetitive thing to discuss. Beside there being a place to talk about it if you see it differently.
It tells you that the fish stinks from the head
I'm not sure 'head technical salesperson at one company became head investor salesperson at another company' is at all necessary to derive that particular and well-worn theory of ichthyo-decay.

If the argument is 'we ought to have a thread dunking on this person', I don't think you'll have to wait long, there will surely be several, as this story develops.

(comment deleted)
The Thomas Midgley Jr of software.
Is the playbook to sell to Intel again?
He’s like the Forrest Gump of failure