When Figma starts designing us

That's true in the world without budget constraints. Sometimes minimal changes in the design result in drastic changes to the implementation effort. Sure, I can build anything, but the client may not be happy if I spend a couple of months rewriting entire front-end just to accommodate designer's dream.

sounds like a gen_redux is needed.

hm ... why not actually? What would go into a gen_redux ... gen_event and gen_statem and some coordinated message passing!

Yep.

Back in the 1990s there was a huge influx of people into web design who knew how to design for print at a "retail" level (design an ad or a poster) as opposed to a "wholesale" level (create a design system for a magazine) and as a dev I would frequently receive a PSD from a designer and figure out how to abuse the primitive HTML was had then to make something that looked like that.

Today Figma has replaced PSD but the same pathologies remain. A new version of iOS comes out and the armchair quarterbacks want to go over the appearance pixel by pixel but they're not really interested in UX design in the sense of designing a sequence of interactions to attain a goal.

As a dev, what I want from designers is design systems, guidance on what everything is supposed to look like that I can implement whatever I need to implement and have it look like a designer was involved. I blame the tools though less than I blame the designers who are just not inclined to think systematically. CSS was definitely designed to create design systems (css classes used in a disciplined way reflective of semantics) but tools like bootstrap, tailwind, Emotion, and the MUI theming system all represent regressions away from that ideal but I don't think those tools make bad designers, it's the other way around.

Having designed websites with Dreamweaver, Photoshop, Illustrator and Sketch, I absolutely love Figma and especially Auto Layout. I don't think people realise how annoying it was to design a list of repeatable items. There already was a concept of "smart object" or "component" that could be duplicated in instances with variations. But laying them out was really cumbersome: if one of the instance had a different height, it would mess up the design because all subsequent items would need to be repositioned. The gap between them was also not a value: it was just space that wasn't used, so you couldn't interact with this gap.

Auto Layout fixes all those issues: you have a list of items of variable height with a fixed gap. You can very easily add/remove/reorder items, without breaking your design. You can even make it wrap, with different column and row gaps, and thus replicate a flexbox layout with "flex-wrap: wrap".

Each item can either hug its contents, have a fixed width, or grow. That's essentially flex-shrink and flex-grow in Figma. So useful.

You'll also notice that prototypes have a "responsive" mode, and it's amazing how Auto Layout will easily adapt to _any_ screen dimension. If you create a data table with one column that "fills" the space, you have a responsive prototype right out of the box.

Also, you can now drag an Auto Layout and it will fill it with component instances and replace its text content, essentially allowing you to fill your design in seconds. Incredible.

If the author still wants to manually place frames around, they still can. Just use fixed dimensions frames, with fixed positioning. That's similar to using "position: absolute" in your CSS. It's just a different type of design. Nothing forces you to use Auto Layout.

Completely agree with the author.

> This is contrary to my belief that any digital design process should start with rough sketches, but move quickly into code and iterate from there.

As a dev, this is the point I resonate with the most.

To me, the ideal dev designer interaction is collaborative and iterative. But the current state of affairs is one where all the design is done upfront, and little is done in terms of explaining why some choices were made. Mockups are not a good medium to spark discussions in the team, because developers are left in the dark about intent.