Hacker News Reader

Comment by pavlov | Hacker News Reader

Designs have to be implemented in something.

Currently UX designers are often like chefs who work by drawing pictures of food and writing down conceptual recipes. When they've created a beautiful menu this way, they're surprised to find out that their recipes can't actually be made using the cooking equipment available, and some of the ingredients don't even exist. You're not going to run a Michelin-rated restaurant like that.

The industry desperately needs better UI development tools. To continue my (already strained) food analogy... Right now we are still in a "hunter-gatherer" stage where everyone does their own thing, teams are foraging and hunting, and occasionally they stumble on something great. Consistent UX will require an "agricultural revolution", a systematic production innovation that makes the essential pieces more understandable and repeatable in isolation.

View on HN

You really underestimate the amount of time it takes to practice and do adequate UX design; unless you spend 10,000 hours doing design, you are going to suck at it, and most of the jack of all trades designers devs I met have really been mediocre at both. "Designers should code" is as fallacious as "accountants should be lawyers". Specialization has been the direction of human civilization ever since the neolithic revolution allowed us to have enough food surplus to move into cities.

Designers should definitely work with devs, if they are surprised their designs can't be implemented in the end, then that organization is REALLY dysfunctional.

We definitely need better UI development tools, but this is orthogonal to UX design, much of which doesn't even involve the aesthetics of the UI anyways.

Designers must understand the media they work in. Design involves tradeoffs and accommodations for practical limits. A designer who doesn't understand the capabilities of the systems they design for is destined to produce bad designs. You can't understand the limits of CSS unless you can work with it.

Imagine if Apple hardware designers depended on their manufacturers to tell them what's achievable with aluminum and glass. Instead Apple has a small manufacturing facility in their design studio so the designers can iterate rapidly and actually understand what is possible.

Yes. But this is far far far away from "designers must code." Except for a few unicorns, this just leads to people who are mediocre at both design and code. A designer who is bogged down in code will also produce bad designs, especially when they let their inability to think of how something will be coded (because they aren't advanced coders) limit them!

Designers have their own tool chain to prototype UIs and user interactions, but these are not UI devtools.

I'm not sure what you're responding to. Did someone say that designers should do all their own coding?

I'll just add one caveat - it's my experience that developers often have little interest in great design code. I write far better mobile design code than any developer I've met (which to be fair isn't many). For me, I contribute best by focusing on that part, leaving developers to focus on their interests which is often new languages, tools, libraries, architecture...

I'm not advocating that designers should code -- far from it. I wish they had better tools for modeling user experiences.

You mention that designers have their own toolchains for prototyping, but these has nothing to do with UI dev tools. I've often wondered why this would have to be so.

Look at other advanced design fields like architecture and industrial design. It would be unthinkable today that professional architects would just draw a facade and not use CAD to actually model the structure. Similarly, it would be very unusual for an industrial designer to model a product using a toolchain that is entirely separate from manufacturing. The assumption today is that CAD and CAM work together, not separately.

Why can't we have the same in software? For my part, I've been trying to work on a solution in the form of apps like React Studio [1]. It's not there yet, but I'm convinced that proper software modeling tools will happen eventually.

[1] https://reactstudio.com

UI dev tools are tools for producing a UI. Prototyping and modeling tools are quite different; it would be unthinkable today for a professional architect to build the building they are architecting and not actually model the structure. Also consider that much of the design that goes on is far away from the actual pixels that show up on the screen; e.g. interaction designers have very little to do with the UI.

> Why can't we have the same in software?

Because devs and designers have different needs.

FWIW - designers are facing the other way - to meeting and exceeding client expectations. In many cases it's not about what can be built, but what could be. It's not great for devs and most of their best stuff gets tossed when it hits real world production. But for a moment, the client was overwhelmed. And that's what sells design.

I would say that knowing some code has allowed me to know when developers are bullshitting me. Most developers I've worked with have been great people, I'm happy to say. But occasionally I come across someone who goes "Uh-uh, can't be done, theoretically impossible, defies the laws of physics" when they mean "I can't be arsed to change the convoluted mess that is our production spaghetti code". Usually at larger institutions. Knowing some programming allows me to call their bluff.

Also, it's far more common that organisational culture, law, or the marketing department is a bigger constraint than the code stack. As a designer, I should consider my "medium" to be far broader than knowing some CSS, or I'm not doing my job.

Being a good designer is inherently being a generalist, and in being a generalist, the designer should understand some basic things about programming in order to have common ground. But that is certainly not the same as saying "designers should code". If designers should be programmers, then they should also be accountants, customer service, marketing, operations, lawyers, c-suite, and every other profession they interact with to solve a problem.

I have spent 10000+ hours doing design and I've got to say that my leap into code (especially backend) has been incredibly eye-opening. I've realized how so much of design is ego-driven, how design studios design for design awards, and what a crime it is that so many technical minds are steered to work on problems more related to fashion than accessibility or productivity.

Designers should definitely code. Too much of design is siloed in fanciful thinking, and fancy people competing with fancy people over fancy things perpetuate delusion.

I agree, but would like to add that there is a caveat to 'designers should code' that is not letting a programming language or framework dictate design through its limitations or structure.

I am a designer who codes, in fact my background is UX and my job is as a developer. It's easy to think about implementation during design, for example, designing a web-app and knowing it will be implemented in Angular leads to thinking about how Angular will implement it, leading generally to modals everywhere, just everywhere. If I knew nothing about the technical implementation I wonder if it would be a smoother design process, possibly better results. But then I see so much poor design everywhere that I can never be sure if something was designed by a coder who knows little about design or a designer who knows nothing about implementation.

Ive spent maybe 1000 hours designing and now 1000 hours coding and i can say actually developing something certainly feels alot more wholesome than designing the layout of an interface.

I constantly have to export my work in webflow for further development on Node, React Webpack, etc. Nothing like fine grained control over the digital elements! :-)

I can do both, design and development. For what it's worth, I can't design the best nor code the best, but I make damn fine products because I'm not caught up in being the best at either.

I make products that ship, people like, and can be further developed in both design and code by specialists. That's good enough.

I agree with the first part! The second part can never happen since there is no process that can handle the complexity of the real world. There might be patterns that could be found generic enough, but not a deconstructed atlas of everything you need to know. It's a process of discovery at best.

To deliver a really great user experience, it's really important to not just look at the interfaces between the customer and your products, but instead try to figure out the "flows" and processes that leads to that the customer uses your platform or app.

In some domains, you will need an awful lot of domain knowledge, paired with UI design expertise, modeling and development skills to do just that.

Typically, it's very hard to find all this in one single person, so excelling in teamwork is always a good trait.

That's a great analogy thanks. I still have the scars of doing the JS powering a crazy UX related to a time-chart zooming in and out with various aggregation level filters. On the other hand, when I told the designers to do X (just change this div to be beveled) they said they wouldn't, I needed to add the divs around it and they wouldn't (couldn't) change the CSS rules to make the UI changes. Sometimes it seems backwards to me, coming from a declarative UI background.

Do you think it's really worth it to have a menu that's so "beautiful" it's difficult to implement without more tooling?

Does it enhance accessibility? Elucidate the IA? Or is it just more whiz-bang?

I'm actually advocating exactly the opposite of "beautiful menus" that are hard to implement.

The current state of tools doesn't allow designers to think in terms of actual UI components and navigation structures. Instead, they are forced to draw pictures and make animated slideshows. This leads designers down a path of tweaking aesthetics because that's what their tools are designed for. (In this regard, newer apps like Sketch and Principle are not fundamentally different from Illustrator or After Effects.)

We should have tools that allow designers to think in terms of higher-level UX components with the confidence that the designs can always translate into code.

I worked on two products that came from this school of thinking, Neonto's Native Studio [1] and React Studio [2] (they're both variations of the same codebase really). I know someone will bridge this gap eventually...

[1] https://neonto.com/nativestudio [2] https://reactstudio.com

Reply on news.ycombinator.com