Hacker News new | past | comments | ask | show | jobs | submit login
Good Design (medium.com/the-year-of-the-looking-glass)
141 points by MarlonPro on Jan 21, 2016 | hide | past | favorite | 77 comments



From gmail's first UX designer:

75% of good UX design is determining in which ways to not be too clever.

https://twitter.com/kfury/status/627350727891906560


When Gmail came out, it was widely panned. It continued to get panned for years for some bizarre design decisions. (let's hide all text labels, let's make compose new email in a popup, let's not let user 'delete' but 'archive'etc, threading... )

Now people don't complain as much, because everyone just got used to it.

Google makes some terrible UX across the board.


Agreed. The Material Design refresh for Google Maps replaced a design that had clear affordances with a completely flat, static-looking view. I'm sure it looked great on some Powerpoint slide shown to the higher-ups at Google, but it was a disaster for actual usability.


yup, and the worst part of material design is, indiscriminate use of bright colors for no apparent reasons. And now everyone is doing the same thing, and 50% of UI look exactly the same.


People eventually accepted the clever and novel UX decisions in gmail, so you know Kevin isn't saying "figure out where not to be clever" for lack of his own cleverness; he's saying it because it's true.


the other 25% is: try very hard not to change anything?


Irony considering the author starts with her experience at OneNote. OneNote (desktop) possibly is the worst designed software in Microsoft Office suite.

I find the following quote from Chris Pratley (creator of OneNote) interesting "You know you have a good design when you show it to people and they say, “oh, yeah, of course,” like the solution was obvious.". No wonder OneNote is so heavily into literal metaphor of a notebook, which frankly does look and feel clumsy inspite they attempting to polish it up a bit off late (it used to look plain ugly till couple of years back).

From official Microsoft document: The user interface is specifically designed to look like a 3-ring binder, complete with tab dividers. Users can divide the notebook into multiple topics and subtopics for the various types of information they want to record and save. Workers have all the advantages of an old fashioned notebook binder with the additional advantages of being able to copy information from one section to another...

Document link: https://www.google.co.in/url?sa=t&rct=j&q=&esrc=s&source=web...


What about that OneNote design is bad? It may look ugly to you but I find it serves its purpose well. It's a metaphor the kind of people who use it will understand.


Well, not answering directly but following illustrates the point:

Apple's iOS7, Well, It Was Time For Skeuomorphism To Die

http://www.forbes.com/sites/timworstall/2013/09/19/apples-io...

One of the bad example of Skeumorphism of Apple http://dun4nx4d6jyre.cloudfront.net/assets/findmyfriends.jpg. Note the 'hand stitched' detailing.

OneNote could have continued to do things well without using clumsy, awkward, out of place design language.


There's nothing wrong with a hand-stitching look. Making an app look interesting is not the problem.

The issues with skeuomorphism occur when the attempt at making a "real life" object imposes restrictions on functionality that would not have existed otherwise. For example:

- Making something look like a "book" causes designers to try too hard to shove things into a two-page view when computers are outstanding at being far more flexible than a book (e.g. multiple resizable columns, scrolling, and so on).

- Making something look like a "dial" just makes it unnecessarily hard to interact with when using a mouse or even a touch screen. Interface elements designed for use on computers are superior.


Thank you. I wish people wouldn't just equate skeuomorphic design and bad design to be trendy (particularly when they don't know what "skeuomorphic" actually means, as often seems to be the case). As you say, there is nothing wrong with having details or styling in a UI, particularly if they are cosmetically attractive, as long as form still follows function.

I think a lot of people on the flat design bandwagon try to create a false dichotomy with skeuomorphism and try to use this as an excuse. All too often, I suspect they really prefer flat design for other reasons, such as not being any good at more sophisticated styling but also not wanting to hire a digital artist onto a project team, or wanting to avoid anything where the implementation requires real graphics instead of whatever severely limited tools are available using nothing but CSS and the icon font from their favourite UI toolkit. I'm not saying no flat design has ever been done well, but all too often I think it's just lazy and/or unskilled design and the anti-skeuomorphism dogma gets wheeled out to try to justify it in some more flattering way.


I suspect the main driver for a "flat" appearance is that basic shapes are easy to implement across screen sizes (e.g. a colored rectangle scales pretty trivially; something like a fancy gradient may require a lot more effort to look reasonable on 14 different device sizes, especially if icon artwork is involved for all the corner and side pieces).

Also, clear icons are easier to construct with simple shapes than with detailed pictures, and symbolic icons may localize better. For example, a highly detailed picture of a mailbox varies between a lot of countries; a symbol of an envelope may not, and it's probably more obvious anyway.

Personally, I think a nice middle ground was the original iOS, where you could do cool things like define simple shapes for tab-view icons but have the OS stylize them for you (i.e. a plain white shape would end up with a cool blue gradient and look perfect in the black gradient tab bar, even though the original icon had no gradients or colors or rounded corners). And the UI elements, while sometimes a bit complex, managed to scale just fine between the screen sizes of the time.


>The issues with skeuomorphism occur when the attempt at making a "real life" object imposes restrictions on functionality that would not have existed otherwise.

Just because you can do something doesn't mean you should. Before relaxing the restrictions on functionality and making everything resizable and customizable, you need to think about the cognitive load that is going to impose on your users. Books are actually a good example of this. I'm not sure I've ever actually wanted to resize or reflow the text in a book. Most books work rather well with the default settings. I can just pick it up and start reading. I don't need to mess with the column layout or scroll settings before I can start doing what I intended to do.

The other advantage that skeuomorphism gives you is that you can piggyback on pre-existing user expectations. A dial may be a bit awkward to use with a mouse, but at least I can look at it and figure out what it does at a glance. The same applies to other skeuomorphic UI elements, like 3-d buttons, notebook-like pages, tabs, etc. I can, at a glance, look at all of those things and know what the expected behavior and interactions are.

Compare this to Apple's flat design or Google's material design. Sure, it looks great. But I don't know how to interact with it. Is this text a button, or is it just a label? Does this panel move when I swipe on it? Does swiping in from the edge do something? How about pinching to zoom? The fact that so many apps have a modal overlay tutorial on first run should tell us that our "flat" designs are unusable. Yes, having some kind of help that explains what each control does may be appropriate, but we shouldn't have to explain how to use the controls themselves.


I've used OneNote for years, but I definitely agree with the parent that it could be better. Here's a brief summary:

Good stuff:

1) Search is fairly fast and comprehensive

2) Supports lots of content types easily with paste, including images and other Office app content

3) Shortcuts for Table manipulation, todo checkbox items, and other icons are really helpful. I wish every editor had these.

4) Handwriting is well supported.

Bad stuff:

1) The organizational structure is overly complex and confusing as it has three categories of structure: Notebooks (top pulldown), Sections (tabs on top), and Pages (tabs on right). This makes it hard to know which to use and it's visually overwhelming.

Wikipedia is a good example of something better as its structure is much more uniform: everything is a page and all navigation uses links.

2) Synchronizing is very slow compared to google docs. This makes it fine for simple scenarios where people edit alone and share after it's mostly written, but makes it hard to work together on a document (no multiple cursors on the page)

My guess is (1) and (2) are an artifact of Notebooks being files instead of "in the cloud". This likely forces the organizational structure OneNote has, and slows down synchronizing because it's harder to sync at the file level rather than the text-run level.

3) Shared freeform content like this gets out of date fast. This should be built into the product so we can see which pages are important to the organization. Most people don't delete pages ever, so some way of showing which pages matter and which should be disregarded would be very helpful.


I also don't find it very elegant but it's the first tool my team actually uses to create some kind of shared document repo.

There are a few shortcomings MS could easily solve but like most tools these days they just fiddle with the UI without adding functionality.

But overall I find it very useful.


What's wrong with OneNote? I quite like it. The only thing that I'm not psyched about is the fact that I have at least three different Microsoft accounts, and all three have separate notebooks that aren't always easy to sync and switch between. But this is a common problem across everything that uses Microsoft account authentication.


We (a group of designers) have been evaluating one note for remote collaboration. Aside from it's egress architectural shortcomings it is non-intuitive in a number of aspects.

'Good design' is sooo much more than the world view laid out in the article. For an organisation like MS, 50% of the job is defending a proposal against 'design by committee' and navigating group politics.

Don't get me wrong, it's a good article but only half the battle, if that!

One note has promise but it's clear that a lot of the design intent got ambushed as the product made it's way out the door.


*egregious

Anyway "intuitive" is not a top priority if you're designing for power users.


It is.

But when you're designing for power users, sometimes it's more worthwhile to trade intuitive design for deeper functionality.

But intuitive design should always be the top priority, IMO - making something less intuitive in order to achieve power-user goals should be the exception, not the rule.

Just my $0.02, and you're free to tell me why I'm wrong. I'm not a designer, I just like using well-designed tools; and more often than not, the best tools behave exactly as you'd expect them to, and communicate intent pretty clearly and intuitively.


But don't you think that even power users desire some system of quick learn-ability? Or do they simply enjoy reveling in their hard-to-come-by knowledge that would deter the average user? I'm hearkening back to Final Cut 7 days here.


Sure, it would probably help. But it's not that important - I mean the product isn't going to succeed or fail on the basis of its intuitiveness. For example, look at all the pinches and swipes that Apple products support. Those aren't intuitive. They made sure to run TV ads for months showing the pinch-to-zoom feature so that people would know how to use it.


If you consider command line as an interface, the number of people who just don't "get it" is huge. POSIX compliance and similar exist for a reason, and yet we have garbage utilities where help, --help, -help, ?, -?, "", info, --info, $1 help, help $1, etc all exist and do effectively the same thing. "Power" users need love, too.

Looking at you, systemd folk.


Yes, the interface has to actually be good even if it's tough to learn.


I have an issue with all the designers who claim to have found the X number of design principles to create 'good' design. There's just something wrong with trying to define anything as 'obviously good'. What's obviously good for me might not be for someone else and the current landscape of incoherently designed interfaces/products/OSes/webpages we use every day just proves to me that if there was such a thing like an obviously good design we would not be in the mess that we call UX/UI design today.


I guess it is like beauty, which is culturally defined. But, as we're heading towards a monoculture (everybody the same OSX MacBook, and thus the same GUI preferences), design preferences become increasingly similar for everybody.

Not for me, however, as I still prefer the CLI in most cases.


Preferring something based on productivity or ease of [power-]use does not necessarily mean you prefer that design.


They're also going with the concept that resources aren' finite.

Between the quest for "the most amazing pixel perfect UX/UI" one didn't have time to properly dispatch NSKeyedarchivers in a queue with groups and synchronise on the main to update the UI or deal with large Core Data writes on a private queue, so now the app is gorgeous but slow as shit until one get some more time.

The problem with most designers who camp in Photoshop is that they become oblivious to implementation details that matters once you run the damn thing.

UI/UX as previously stated, is very subjective although there is always objectivity to performance. But hey, at least we got that amazing custom menu bar that is totally better than the other menu bar that wasn't exactly the same.

Those who've worked in body shops will know this. Take a look at the new Twitter client. Obviously it has some nice style to it. But it's slow, buggy, and crashes.

Factor in that resources are finite and design can be good enough, but performance can't be good enough.


I disagree. UX/UI design has to be obvious, people need to know what to do, without having to ask for help or look at a manual. And 'obvious' is easy.

Certain arrangements that already exist are basically universal. People know where to look, people understand certain symbols. Use those, don't try and reinvent them. Use what people know.

Trying to be clever is what dooms design.


Rocks and sticks are simpler than computers. By your logic, we should still be using rocks and sticks.

Learning is a necessary part of getting efficient with complex tools.


Trying to be clever is the main thing that has led to progress and eventual better design.

Amazon's drop down menu comes to mind - http://bjk5.com/post/44698559168/breaking-down-amazons-mega-...


Ah, I think you missed the point. Amazon's solution to the menu behavior, yes, is clever. But what it presents to the user is not. It is a dropdown menu, that looks like every other drop down menu ever made. It's not novel at all in that respect. A user sees it, and they instantly know what it is, and how it works.

Yes, Amazon did something clever to make it work as well as it does, but barely anyone even notices that there's anything special about it. It 'just works'.

That's key. The actual design pattern here didn't change at all. It just works better.

Can you make something work better? Do it. Are you trying something because "icons and folders are a dated idea"? Then don't.


Fair enough, but then wasn't the idea of icons and folders considered clever at some point in the way you are describing. It seems that the vast majority of design is not better than the current accepted standard, but every so often someone designs something which is good and becomes the new standard. At some point you have to be willing to do something potentially stupid if you have any interest in pushing the boundaries or am I still missing the point you are making?

I think gmail's redesign from a few years ago falls into what you are describing - trying to design something more "modern and clever" which in some way was to the detriment of the users (at least if you were judging by the reception and all the critique). But at the same time if you don't try to maintain a more modern interface I think you risk being left behind, so you have to balance "clever" with "expected" design.


What is your intended goal in 'pushing the boundaries'? Where is the reasoning? Are you making life better for people, or are you confusing them? What is 'falling behind' in design? Does your product work well? Is it easy to use?

If you believe something has to change, just because it should, then I think you're coming from the wrong place. You have customers that depend on you, your job is to serve them, not some sort of need to invent the next great design revolution.


Most problems already have some sort of solution for them. When you make a claim that your solution is better than the existing solution you often provide a better UX in some respects. For this you have to make design decisions some of which don't have an accepted standard. This is where you are pushing the boundaries.

On the other hand, UI style and UX naturally improves over time. Compare Windows 95 and Windows 7. There are certainly style and UX changes in the latter that are superior. A lot of people may not like Window's Ribbon interface but I think it was definitely an interesting new design, one which I think will in the long term prove to be better than plain drop down menus.

So I guess I am making the claim that if you are really invested in your product you probably have ideas on how it could be improved and a lot of those ideas are uncertain but deserve time to be attempted.


There can be obviously good principles while design still remains hard. Chess, for example, has plenty of simple principles (like "control the center of the board") that are hard to execute.

However, designers writing articles tend to be looking for insights that advance their knowledge. They don't write about things that are obvious to them, even though those are the most useful for the rest of us. And most non-designers are really looking for typography and color selection advice, which has a similar relationship to design as programming has to computer science.


I think it's ironic because it's on medium and the first thing that comes up is a giant, stupid, unnecessary picture that is full screen and could lead people to believe that there's nothing more to the article.


This: "Better design does not mean more design. Often, the most obvious designs are invisible." This also reminds me of the Steve Jobs approach of being proud of all the times that they said "no".


Too bad he isn't around anymore to say that to the Apple Watch user interface.


Or, the iTunes radio app.


Or, any version of iTunes after, like, 7. :(


This article is a clumsy and self-centered retelling of "Dieter Rams: ten principles for good design" from 40 years ago.

https://en.wikipedia.org/wiki/Dieter_Rams#Dieter_Rams:_ten_p...


Looking like being obvious is not among Rams principles.


#4: "it is self-explanatory."


OK, I think it's a misunderstanding. Self-explanatory refers to the use of object that is designed. It needs no explanation. How to ride a bicycle is pretty self-explanatory (even if it's not trivial to learn).

The design of a bicycle however is not obvious.


+1


What an overly simplistic view of what makes design good (as if that can even be defined).

Taking any advice from a designer at Facebook is a dangerous game to play. It's very likely you work for a company that is structured entirely differently than anything close to Facebook. What makes design "good" at your company is going to depend wholeheartedly on the market you serve (which is not going to be 1 billion people large).


This made me think of "The Making of: Dust", which describes the iterative (and serendipitous) process of creating [two of] the most well-known Counter Strike maps. I think it's a brilliant how a design that feels obvious involves a lot of work and uncertainty.

http://johnsto.co.uk/design/making-dust

HN: https://news.ycombinator.com/item?id=9772521


Here are some more abstract tips that might apply to more than just Facebook's mentality.

1) Have a clear workflow that matches your business and your user goals.

2) Make it as short as possible.

3) Don't have dead ends.

4) Make sure 4/5 people in your target market can jump into it and accomplish the task within a few seconds without asking any questions. (This is a good target for a V1 release, V2 and beyond should reach for 9/10 or more)

The rest is mostly trend and opinion. Good design is just good HCI -- fast (both quick and understandable) interfaces to complete a task.


Loved the article, but I don't like Julie's redesign of Facebook. Am I the only one? To remind yourself, here's pictures of older facebook designs: http://getlevelten.com/blog/julie-miller/if-you-didnt-design...

I prefer the 2010 design by a mile. FWIW, I also prefer Gmail's old design.


I think the current photo-centric design just reflect's change in people's behavior shaped by growth in mobile phone usage.


How has the timeline changed? I can't remember the last time I actually looked at someone profile to be honest.


Good design: Then we are taken to a page with a banner so big that you have to realize that you need to scroll way past the bottom of the screen to actually reach the content.


"Obviousness comes from conforming to people’s existing mental models.". It's currently pretty common to scroll webpages with big images in the middle.


Well, it still doesn't make it "good design".


"Obviousness comes from conforming to people’s existing mental models. Don’t waste time reinventing common UI patterns or paradigms unless they are at least 2x better, or you have some critical brand reason to do so."

This is a beautiful insight. I wish everyone understood this. There's ZERO reason to reinvent the wheel with design.


Some decades ago some architect said "why do we need a facade? Isn't it redundant?".

Noww I can't find main entrances to buildings. Please stop this experiment.


I can't even figure out how to get on the property sometimes - see the old EDS "Mothership" in Plano! [1] Amazingly, the front desk was on the 4th or 5th floor archway - which you could not get to without getting past a badge reader. (that's the way I remember it anyway)

Street signs, entrances, etc should be optimized for both first time users and everyday users; which is difficult, but without that traffic goes slower and you have more accidents, people get frustrated and hate you.

[1] https://h71044.www7.hp.com/campaigns/2011/events/POD/images/...


> Better design does not mean more design. Often, the most obvious designs are invisible.

I think "design" is ambiguous here. You can look at it as meaning "effort that went into creating something", where "design" is the thing that designers do at their desk.

Or you can think of it as quantity of "stuff" in the end product that a person experiences and consumes. Every button, doo-dad or blob of pixels in an app. Each nut, bolt, accessory or function in a thing.

In that sense, my personal definition of good design is doing more design work to deliver less design stuff. A designer's job is to spent time chewing up and swallowing complexity so the user doesn't have to.


I agree with this in principle but today's so-called designers are enamored with minimalist designs to user's detriment, relying too much on white space.


The article feels ...well 'obvious'... but without any real insights into what good design really is. It's a little bit like saying "What's a good investment" - One that makes you money. Or "What's good code" One that doesn't have bugs. It tell you something that's obviously true, but there isn't much you can do with the info.

Compare that with a nugget of wisdom from Charles Eames - someone who actually built things that designers can look upto. Humble, to the point and not obvious but insightful.

Is there a design ethic? There are always design constraints and these usually include an ethic.


I feel that rule number 2, "If you cannot get a group of people for whom your product is designed for to generally agree that your design is good, it’s not good," is generally a double-edged sword. That is, do people outside of design generally know what is or isn't good?

Rule no.3 stating that a group of designers must decide if a design is good or not is far more relevant to what actually is good design, verses the general populace.

That's like asking a group of patients if they think the drug that's being made seems good enough with several side effects or not, verses asking pharmaceutical experts on the matter.


I think both points of view have merit, but the user pov should probably be more heavily weighted.

My experience with user testing is that if you give someone a UI and watch what they do with it (without instruction), you will quickly get a feel for how close to the mark your design is. Different users have different expectations of course, so you need to do this with a fair range of users.


True, and I do think there is strong validity in blind user testing so as to better shape and improve initial designs to a more finished form.

However, I do think there is some merit in getting those in the actual field of functional and "good" design (as the author defines it) to weigh in on why precisely your design is just that. In my own UI/UX research, users all too often miss fully comprehending the painstaking research and work that goes into truly effective and good design.

I guess what my argument really is is that I would prefer to weigh more heavily the affirmation of those in the design field rather than a broad range of users with no particular ethos concerning design. Of course, who we end up designing for is the normal user, and what they think is indeed important. However, if 100 people said my design was good vs. hearing that from, say, Mike Matas, I would pick Mike Matas.


Users cannot tell you what's wrong with your design. They don't know. It "feels wrong" or they "don't like it". But they CAN tell you that your design is BAD. It's up to you (or other designers) to tell you what's wrong with it. But if the users don't like it, you failed, and it doesn't matter if other designers think it's good.

Your target is the users, if the users aren't having a good experience, you're not doing it right.


I agree. Users loving a solution is much more valuable than designers loving a solution.


I think a better analogy would be if a chef made you a meal and asked if you liked it. He is the expert but your experience about the taste of the meal is still valuable to him.

Testing on your target market is pretty important because you get a "real" example of how usable your product is. For example if you tested your interface on an elderly person (target market) and a UI expert you are going to have considerably different results.


Such a qualitative view of design--Nothing about measurement or business goals. I would argue that one design is better than another only if it can be shown to improve whatever business results you're looking to improve. Are you trying to increase 7-day retention? Get more people through the sign-up funnel? Increase conversion? Test the candidate design next to the old one and see if it's actually improving things. "A bunch of other designers think this is a great design" is not really good justification for spending development hours implementing it.


Good luck testing it without developing it. And double good luck getting something without investing in it. Apply business process after you actually have the kernel of a product, which includes a product design.


These principles extend beyond commercial product design. The API and thus the ergonomics of a programming library come to mind, for example.


A/B testing is definitely a way to go - when you have enough users! And it only helps you decide between solutions you already have.


The idea of that "good design" is something you "just recognize" is somewhat related to the ideas in the book "Zen and the Art of Motorcycle Maintenance", but replaced with "Quality". Really good read.


Good design is when your hopes and expectations come true

meaning: When you do a user action, like double clicking, the application should do what you expect and hope it will do.


Lots of focus on too many restraints, but I'd say poor design can just as often be the result of not enough restraints.


After reading this, I went through and read all the other linked essays she wrote. Good stuff!


Great post. Thanks for sharing. The same way a great song sounds familiar the first time you hear it, good design feels comfortable and effortless from the first second.


Great article, definitely a good mantra to keep in my head when I struggle against muggle resistance...


Like most articles on medium: self-bluffed nonsense...




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: