twopointouch

This is a personal attempt to understand the digg front page. I am not a mathematician, nor a coder nor an Excel wiz (all of which will become obvious). Nonetheless, I wanted to understand digg better than I did and decided a tiny bit of analysis was in order.

This was the state of play on the front page at 16:25 BST today.

(Time means the number of hours since a post was submitted, to the nearest hour.)

Things to note? Your ranking on the digg site does not correspond any to the three metrics available to users. Nor any combination of them that I was able to concoct. I tried a lot of formulas, but nothing that worked. If the number of diggs was key, then the number one story would be on page 3 or something. All the front page stories have attracted a fair amount of comments, but there’s no correlation between that and their rank. Comments are good and diggs are good, but there’s no way of knowing how much. I could post some graphs at this point, but since there’s no visible relationship, there’s no point. It’s clear that each story has a time-to-live on the front page, but impossible to detect what that is.

However, we’ve learned that digg front page stories do appear to be submitted within the last day, with around 15 hours being the average between submission and the front page.

It makes sense for digg to use a secret algorithm for posts, one that isn’t easily available to users through any of the information they’re given. Otherwise, the solution for gaming digg would be publicised, get spammed and the service would lose its users.

What we don’t have is either the rate of diggs or the rate of comments. I think it would be fair to assume that the ‘G-Meter in 1 minute’ story, the number one story in the list at the time I recorded, gained either a large number of diggs or comments over the recent past. Otherwise, the number two story, and all the rest, would have a higher place. Rate of diggs or comments seems to be more important than their number, although both are important. Since this is about just one moment in time, I can’t comment on how quickly or slowly a post rises and falls dependent on that. However, this is a key indicator, I am sure.

So I failed to reverse-engineer the digg algorithm. I’m frankly not up to the job and I don’t believe we have the information available. So then I looked to a more folksy way to understand the page. Folksy is my forte. I tried to come up with tags that would categorise the stories I was seeing. I know this is not academically rigorous in any way. But this was how it panned out according to my own categorisation of what came up:

So all of these things are good. Again, it’s just one moment from the contantly circulating digg nexus. I wouldn’t want to draw too many rules from this that you couldn’t work out for yourself. Web, Major Vendors (especially MS & Apple) and Conspiracies seem especially good.

I look forward to future posts about ‘How To Beat Microsoft’s Planned Web Conspiracy About Melons’.

P.S. working out how ‘Friends’ or voting blocs might contribute to any of this is well beyond my reach. Alex has some thoughts on this. Thanks to David for the link.

P.P.S. Maths geniuses are welcome to the data here. (.xls file)