Noam Rosenthal

A #webperf thought after profiling some sites in the wild:

Look at the different parts of the DOM tree & JS that your users get. Do they look like...

1. Your mental model as a dev
2. Your org-chart, how your teams are set up
3. How users experience your site

If it's not (3), perhaps your tools are wasteful?

Dec 18, 2022 at 1:49pmWeb
Show threadRyan Townsend (old account)Dec 18, 2022
@noam you missed 4. a jumbled mess that's so far separated from the code you write that you've little hope of understanding how it'll really act or perform in production based on your local view 😰
Show threadNoam RosenthalDec 18, 2022
@ryantownsend point taken! Though often in these cases I wonder if your org is also a mess and this is really (2) 😈
Show threadRyan Townsend (old account)Dec 18, 2022
@noam it doesn't take more than one developer doing resume- or FOMO- driven development to unnecessarily adopt an SPA framework and complex build tools.
Show threadNoam RosenthalDec 18, 2022
@ryantownsend ok that would be (1) - your users get the DOM and JS that look like your FOMO and response to hype
Show threadRyan Townsend (old account)Dec 18, 2022
@noam ah yes, I misunderstood that point – I was reading "mental model" more in terms of code structured based on a limited understanding / non-user focused view of the problem, rather than bringing emotional baggage along for the ride!
Show threadPatrick MeenanDec 18, 2022

@noam
There's something between 2 and 3 that feels like tradeoffs between DX and UX.

We need good platforms/frameworks that allow for org-level separation of responsibilities while not penalizing the UX.

Feature velocity and independence will win out over optimized output for the browser 9 times out of 10.

Show threadNoam RosenthalDec 18, 2022

@PatMeenan

Absolutely.

"...platforms/frameworks that allow for org-level separation ... while not penalizing the UX" is exactly what I'm interested in.

Show threadNoam RosenthalDec 18, 2022

@PatMeenan btw I wish we could somehow measure "how much UX latency does your DX cost" - i.e. what would be the difference in perf metrics for the same UI without any DX "tax".

If we could measure that, people could have better visibility of the tradeoffs they chose. Right now it often feels a bit opaque.