I was listening to a podcast by tech legend Leo Laporte in which he interviewed someone on the Adobe Photoshop team. This was a big deal because they were talking about the new release of Photoshop, which is big news because nearly everyone who has anything to do with graphics has used / probably still uses Photoshop. One of the things they talked about were speed improvements. In all past versions of Photoshop, it ran much, much faster on Windows PCs than on Apple Macs… and this was something Mac users were ashamed of… so they didn’t bring it up often. This was true for all PowerPC-based processors, including the G5.
Now with the release of the new universal-binary Photoshop, it should run on Intel Macs with similar performance to PCs… in theory. They actually don’t know, and the benchmarks haven’t been done yet. The point, though, was not the relative performance… or even the actual performance at all. The Photoshop developer said that they had done many things to improve the speed of the product… but actually the perceived speed. That means how fast the application feels… not how long it actually takes to do things.
This is disconcerting at first, especially to me, a person who likes objective statistics… but it’s really true. It doesn’t matter how fast your algorithms are if the users cannot feel the difference. In fact, it is this perceived speed that has the greatest influence on user satisfaction. I’m sure Google understands this… that’s why their search results are always returned in less than a second, and they’re buying up dark fiber to ensure they have fast lines directly to your PCs and other devices.
So I was surprised when I suddenly realized while using Blogger today that Google’s blogging service– Blogger, formerly of Pyra Labs– was painfully slow in terms of perceived speed, at least if you’re publishing to your own server (or a non-Blogger-hosted server). You actually have to sit there and wait while Blogger logs into your server via FTP and publishes all the updated static files. This is good for sites with high traffic… no dynamic content means the server can handle a lot more traffic at a much lower cost. But you suffer in terms of the perceived speed of publishing.
This isn’t the way it has to be, of course. I don’t know if Google realizes this, since these are acquisitions… but a big part of YouTube’s popularity is its speed. You wouldn’t want to sit there waiting 1 hour for a video to download. And similarly, you wouldn’t want to wait an hour for YouTube to convert your xyz-format video. The thing that amazed me about YouTube video uploading when a tried it a few weeks ago is that it doesn’t make you sit at its webpage to wait while their computers are processing your video. Instead, once the upload is complete, you’re free to do whatever you want, leave YouTube, watch some other videos, etc. The upload experience is relatively fast and simple… despite the fact that converting the video takes a long, long time!
How do they do it? They have a separate process running server-side, so right after you upload your video, it’s not online for quite a while. This was frustrating to me at first… until I realized that this is a technical necessity because of the computing hosepower required to manipulate video (and many different video formats, at that).
My Blogger feature request, therefore, is to do something similar with Blogger. Instead of making the publisher wait, just say it’s done right away, in less than a second. And leave us to go do something else… write another post if inspiration strikes us again. Blogger would be logging in to the FTP server in the background at its earliest convenience… but now it doesn’t have to happen right away, and we don’t have to wait until it’s done.
I considered possible downsides to this model. Well, it might be technically hard to implement, but that’s no excuse for a company like Google. It might mislead users because they won’t be able to view their published post right away… but they’re not actually waiting any longer for it. It’s just that they can’t see when it’s done. So Blogger would ideally have a small, unobtrusive status indicator at the top of pages… and it will tell us when the “live” version of the public blog is up-to-date.
This is a really good solution, I think. Blogger team– are you listening?
Update (September 19, 2007): The new Blogger fixes this problem by doing custom domain hosting. Thanks, Blogger team!