list at joreybump.com
Fri Aug 20 17:34:21 EDT 2004
Scott Sanders wrote: > I am not deluding myself. If you have two pages that take 50ms to > render, but one is streamed, every user on this planet (including > yourself) will think the streamed page is faster. It is a psycological > fact of life. That is part of the reason there is a computer field > called HCI. > > We had a customer that was pissed off about how slow our server app > was. All we did is turn off buffering of output (we were doing it to > catch errors before the user saw the page), and the customer said > 'That's great! Why weren't you doing that before'. Render time AND > response time were the same, the only difference was TTFB (time to first > byte). I'm not questioning the practice; indicators are a good thing. But in spite of the initial rendering, spinning hourglasses and whatnot, users are fully aware of how long it takes to get the information they want. When it comes to users, "slow" is a vague term; "broken" is not. By providing indicators, the user gets a buffer in which they don't need to worry that the app is broken (yet). This may be the relief you offered to your customer by decreasing the TTFB.