I really like the distinction of "density in time".
JIRA is a really visually dense application, but it's speed, as well as the number of different screens you normally need to click on makes it feel really sparse despite the dense visuals.
JIRA is a really visually dense application, but it's speed, as well as the number of different screens you normally need to click on makes it feel really sparse despite the dense visuals.
[0] - https://blog.codinghorror.com/performance-is-a-feature/
A decent level of performance these days should be table stakes, and _high_ performance is a feature. Software that’s molasses slow _is bad software_ these days.
And when slow is the only option, people will wait.
Decent performance is table stakes outside of very special contexts, and software that can't manage it is bad.
Finding an exception doesn't make that stop being true in general.
Stable diffusion is not relatively slow, because it has no alternative that is noticeably faster.
A great example is YouTube in a web browser. My internet is 350 Mbps with 20-40ms latency. Trying to load YouTube in a new browser tab takes a few to several seconds and I'm forced to wait for it to load because the sign in link doesn't show up until the end of the seizure inducing re-render flashes. Safari, no add ons.
I can't believe it takes so long and I think less of Google as a company because of it. Them speeding it up is not a feature in that case. A trillion dollar technology company ought to provide fast as merely baseline. Anything less is them intentionally disregarding their customer.
Also 1s is super slow given how much time you spend in the tool and how many clicks are required for many operations.
I’m looking at a Kanban board right now. I have a 41 inch display and I can see a total of 9 issues at a time across three columns. And that’s with basically doing everything I can to maximize the space for the board and minimize how much chrome goes around it. I have no idea how anyone uses this thing on a laptop display. It’s awful.
Differance - the difference that makes a difference.
It's not just more, or density: you need to sum the cost of context switches, which depends on context size and continuity with prior, i.e., delta size, which in turn depends in part on relevance i.e., which parts really matter.
Design by committee (including one person over time) loses the natural continuity and integrity of an initial idea.
The first derivative of difference?