Yes, title is kind of a homage to that false dichotomy posted yesterday by Spolsky.

Yes but StackOverflow is very similar to Digg:

you've got articles that have rating, you've comments to those articles that have rating, articles and comments are sorted according to rating, people earn karma from posting good comments / articles, everything has tags.

So personally I don't see a false dichotomy; even if Digg is more dynamic / complex ... WTF are they doing with those 500 servers?

Digg is more like Facebook and Twitter than StackOverflow. Each of Digg's logged in users gets their own "News Feed" based on the users they follow/friended/are most similar to.

StackOverflow on the other hand is much simpler - questions and responses, plus users and voting. AFAIK there's no collaborative filtering going on at SO, be it user or item based.

I don't know why Digg needs so many boxen, but I did find Spolsky's comparison disingenuous.

Well TBH 500 is ridiculous overkill. Even 100 would be pushing it.

Whether it's overkill or not depends entirely on what the servers are being used for. Without actually working there it's hard for one to say for sure.

We know what the service does. We have that much data - we can see what users are allowed to do on the website, and what they cannot do.

And for the level of content, interactivity, media, etc etc, I can say that 500 is silly.

You don't know what each server is doing. They might have 10 servers being used by an internal marketing analytics team, with 40 support servers for development, QA, testing, and disaster recovery of those specific services.

What sort of computing resources do you think it took for Google to develop the autopilot car? Would you have been able to determine that by looking at their homepage and the services they provide? No. That's the point. I'm not suggesting that Digg is doing anything so interesting, but most of those 500 servers are almost certainly NOT being used to support their website directly, they're being used by the business for other things.

The title was supposed to be a serious representation of our traffic numbers (our amount of servers wasn't supposed to be part of the comparison, the poster here is a slippery snake) but we can pretend it was if you like, nobody likes joel so it fits in well.

I liked this submission simply because it demonstrates the ridiculousness of such claims.

Spolsky compared two rather more similar sites though. This one is a wiki where most of the content is static, served out of cache. Still, I agree that such comparisons are on a very weak footing. A single feature could easily kill any valid comparison.

Thankfully it wasn't actually less traffic, it was only 4x (maybe 5x) the traffic, with 10x the servers.