It's not just everyone playing Scrabulous that's making Facebook so slow -- a new study suggests that social networking sites rank pretty poorly when it comes to overall performance. But can its results be trusted?
A study by performance monitoring company Watchmouse assessing site loading times for 104 social networking sites found that most had fairly poor performance. Facebook, the poster child for how to build a social networking business from nothing, got the worst result of all, with a score of 6629.
Under the study methodology, a score below 500 indicates decent performance, while anything above 1500 is a major problem. In this context, and if you believe the numbers, Facebook is performing about as effectively as a Rubik's Cube that's been repeatedly dipped in fast-setting glue.
In an interesting demonstration of the broader power of social media, the list of sites was drawn from a Wikipedia entry on the subject, rather than applying any more specific criteria of relevance. That means that a huge number of the results apply to sites that, frankly, no-one's ever heard of, including the half-dozen sites that actually managed to get scores below 500 (Flirtomatic, anyone)?.
Amongst the better-known brand names, Yahoo! 360 (707), Bebo (912) and MySpace (923) didn't do particularly well, while LinkedIn (1006), del.icio.us (1049) and Windows Lives Spaces (1386) all crossed into the dreaded four-figure zone. Along with Facebook, Last.fm (1837) and TagWorld (2644) were amongst the better-known sites getting an official black mark for scoring above 1500.
“It is interesting to see that popular networking sites turn out to have very bad performance," Watchmouse CTO Mark Pors gloated in a release announcing the results (perhaps he's hoping this will drag in a few customers to utilise the company's commercial service).
"It is surprising they still have such a big fan base when they serve their users so badly. Using Ajax technology, they should be able to work more effectively. For now the sites will need to do a lot of work to remain popular and improve their performance."
But can we believe it?
An examination of Watchmouse's methodology suggests that it may have been both generous and cruel with these performance figures. Watchmouse derives the figures by averaging the time taken to access a given site's home page, without frames or graphics, over 1000 attempts.
The home pages of many social networking sites are astoundingly basic -- Facebook's consists of little more than a log-in box and some basic text promoting the service, so it's not exactly Ajax Central, and its outcome might have been worse if other pages had been taken into account.
At the same time, not examining a range of pages from each site arguably distorts the typical user experience. MySpace, for instance, emerged in the middle ranks of the survey, with a score of 923. However, given that MySpace users can embed all kinds of complex (not to say browser-breaking) code in their profiles, examining a range of pages would almost certainly have produced a lower score.
Facebook's status as the most-hyped and fastest-growing social network might also have ruined its figures. Watchmouse applies a 10-second penalty if a site fails to load because of errors on its test machine and a second check machine. Almost three-quarters of Facebook's performance score comprises this kind of penalty data, rather than actual load speed on successful attempts, and if the test got done during the recent Scrabulous server update, for instance, that might have artificially distorted results. (Watchmouse gave no indication of the timeframe involved in the study).