Why computers need a long start-up time
I had to look something up on the interwebs this morning, but did not really feel like firing up the old PC. That’s really what it feels like: waiting until the information service at my finger tips deigns to service me. Even the fact that Firefox takes seven to ten seconds to start up doesn’t bother me, because that is peanuts compared to the time the computer needs to get ready.
Google itself? Less than two seconds. Not instantaneous, but good enough.
I doubt a long start-up time is necessary. There have been experiments years ago with keeping part of the on-state on, so that the PC only needs to load a couple of things while starting up. And of course, lots of people nowadays just let their PCs sleep, and never switch them off. I would too, but the noise from sleeping Macs does nothing to convince me that these things are hardly using up energy, and my sleeping PC (quiet as a mouse) has a tendency not to wake up.
In his excellent Don’t Make Me Think, Steve Krug describes one of many problems that besiege a homepage; everybody wants a piece of it. This is literally what used to be happening to the homepage of the city of Amsterdam; the many departments all wanted to have a story on the homepage that led to their department’s website. If the webmaster had given in to this, the homepage of www.amsterdam.nl would have been unusable.
My guess is that this is what is happening with a computer’s start-up sequence: every stake-holder wants a part of that pie. Even if the producers of the OSes I use could get their start-up sequences to under a second (they can, but they won’t), programs I install would still be able to ruin it by claiming part of the start-up sequence.
Only devices that aren’t viewed as computers have a chance of having a limited start-up sequence, because their users wouldn’t stand anything else. Can you imagine having to wait thirty seconds before being able to answer a call on your mobile phone? The caller would have hung up by then, or your answering service would have kicked in.
(So why do we accept this situation with PCs? As Steve Krug says in the sample chapter I linked to, “we muddle through”. Start-up time would be an interesting thing to have fixed, but in the absence of a fix we accept the lesser solution.)