When design doesn't matter
medium.com
medium.com
But I think you've pointed out something poignant: for many, design is "making something look good".
Function is fundamentally undetachable from form, so a poorly designed product can't truly be said to look good. It can just have varying degrees of fitness.
I highly recommend reading "Notes on the synthesis of form" by Christopher Alexander. It might blown your mind on what you consider design. I feel this word is tossed around a lot but few realize what it can entail.
The idea that a poorly designed product can't truly be said to look good seems subjective. Could you elaborate? (Especially on how this might relate to digital design?)
It would help if you had read the book, it's the kind of thing that "snaps", but I'll try to condense it:
Think about forms as physical manifestations of the underlying forces acting on it. A good example I once heard is thinking about a water drop: if you take a picture of one, you can explain why it has the shape it has because that's the result of all the laws of physics acting on it at that moment. And because of that, it really couldn't have a different form than it has, unless you bend the laws of nature. It's the purest form possible.
Analogously, a human-designed artifact is a physical (or otherwise) manifestation of the underlying forces acting on it (which would be the requisites). In a concrete example, if you were designing a kettle, that would be all the things you need to optimize for: it should be placed atop a oven; it should have such and such size to fit in a cabinet; it should allow water to heat; it should allow someone to grab it without burning their hands; and so on.
As you can imagine, it is really hard to enumerate all the required (and often conflicting) requisites. For instance, if I were designing an artifact for heating water without stating that it should be placed atop a oven, I could arrive at a completely different form (e.g.: faucet heater). The obvious takeaway is that design, as a creative discipline, is only as good as the number of requisites a designer manages to contemplate and successfully meet (let's call it the "fitness function").
Therefore, for all collection of requisites (underlying forces), a perfect artifact with 100% fitness should exist (purest form). If that's true, then good design can be judged objectively by its "fitness function", quantifying how well it meets the proposed criteria, or how far it steers away from the "purest form". Better yet, if we devise an ideal design process, these optimal forms themselves could emerge from the requisites (that is Christopher Alexander's research, and from there it gets heavily philosophical).
To wrap it up, when you say an UI "looks good" at first glance, you are mainly judging the artistic prowess on it (which is subjective). The real judgment of the design though is in how well it meets the requisites (fitness, or how far away it steers from an ideal UI). So when you use said UI, you are judging the design when you quantify: how intelligible it is, how well it allows you to do what you need to do, and so on (since those are the raisons d'etre of UIs). And yes, eye candy can very well be one of these criteria. In fact, that's the criteria most products are optimized for, subconsciously, mainly because it's the easiest to quantify (designer shows the design to client, gauges emotional response, iterate). Hence why our confusion between design and eye candy.
Based on this definition of fitness how would we evaluate the design of Craigslist, Comic Sans, and the Drudge Report?
Fitness would be a function of how well the realization maps to the problem domain. That is, how far these are from an ideal Craiglist, Comic Sans and Drudge Report that would each fulfill all the required criteria.
It's impossible to evaluate detached from a collection of requisites. That is the hard part. Christopher Alexander proposes a (mathematical) method for collecting these requisites and deriving a design from it by building a direct graph, study the relationships by grouping it, then realizing it by building in successive layers, but not the reverse (taking a realization and working your way backwards).
You don't need your web hosting provider to look good on the backend as long as everything you need is there and fairly easily find-able (and that itself is good design). Doesn't matter if they provide an awesome service if you can't set it up and use it quickly and efficiently.
You also want your front-end consumer-facing application to look good, especially if you're trying to disrupt a market rather than create a new one.
I think a good product embodies just enough design + engineering to make something useful.
The risk, specifically with visual design, is that we can "over-design" (in the same way that we can over-engineer).