Establishing stats

Headlines today say “Silverlight now on 60% of Internet-connected devices” and “MIX10: Microsoft Silverlight lands on 60 percent of Internet devices” and such.

But the sourcing is left blank… some guy on a stage said so, and so newspapers are repeating it. That’s not good journalism. Anytime you read any story, you have to ask yourself “How do I know what I think I know?”

I know of two services which measure Silverlight: Travis Collins’ riastats.com, and StatOwl.com. RIAStats went through a significant sampling shift towards the end of last year, roughly doubling its baseline levels, and now reports SL3 at 53%. StatOwl shows SL3 at 38%. (Because of default auto-upgrades, I’m assuming any remaining SL1 or SL2 installations are broken.)

If an exec stands on a stage and gives you numbers to repeat, that exec has a responsibility to show why you should believe those numbers, and how you can defend them from challenges. Repeating those numbers without curiosity does not do the planet any favors….

If they can get Silverlight up into Shockwave/QuickTime levels of consumer support, then mazel tov… now that all this “Flash-killer!” adversarial jive is behind us it’s easier to just wish them success in their goal of helping skilled .NET coders deliver to wider audiences.

But — the newspapers need to source their info. Otherwise they’re just giving us beliefs. “How do you know what you say you know?” must be asked.