Finding UX in the Trash
f2.svbtle.com
f2.svbtle.com
The things you learn talking about design ;-)
Firstly, it's hard to put out. Even though there's lots of frozen water, liquid water that you can throw onto a fire to put it out is lacking.
Secondly, if a small structure catches fire, it's not like in the middle of a US city where you can just stay at a hotel and buy what you lost. Depending on the time of the year, the tools inside might take months to replace, not to mention the shelter itself being lost.
Nothing is more infuriating than finding out that something that should be straight-forward to do is hard or even impossible because the option got taken away in the name of UX.
Actually, there is one thing more infuriating: If that UX was also inspired by business goals and not user interests.
Being opinionated is fine, and for the most part Apple makes the right calls. But it’s not as simple as dictating “thou shalt...” via the UI and calling it a day, and then flatly refusing when questioned. It’s the UX equivalent of a bad manager who only learned the “be an asshole” from Steve Jobs-style leadership while forgetting all the reasons he could get away with that.
In a properly modular design, you often do not need to choose whether to upset the expert user or to make the naïve ones happy.
Like.... Facebook ? I do not use it myself but ended up checking somebody else's page for a specific comment. Took me a while !
I remember a while ago (4-5 years ?) there was some fuss about Facebook deciding not to improve UX because tests showed users would spend less time on it. But cannot find anything at the moment.
Facebook is the digital equivalent of that.
The milk is in the back on purpose, but products are grouped relatively sensibly too.
The design of the path through the storey was not chosen to annoy you, but to make optimal use of display and saleroom space.
But yes, Hilbert curve rather than maze, makes sense.
The advice in the article works well for simple things. Simple things can, and should, be perfect.
Unfortunately not everything can be simplified until it makes sense to an unskilled operator (or an unskilled designer). Ideally, complex things are composed of simple things... but if that were true, wouldn't we'd all be running UNIX command-line desktops?
Contrary to the author's leading premise, the best designs don't just work by taking options away from the user. They hide those options until they're needed... and when they are needed, they make those options easy to find.
And that's hard work, the sort of hard work that can't be waved away with a list of trendy bullet points from some respected design guru who mysteriously never seems to be around when you need to use a different printer than the one the OS thinks you want.
That seems backwards, as the argument being put forwards is that UX is all about trading off a complex, options-rich interaction with a simple, options-poor interaction. So good UX delivers simple things. The hard work comes in the skilful reduction of a potentially complex interaction to a simpler one, and recognising that there are tradeoffs.
Sure the Mac printer selection experience is great when you have a laptop that you carry from work to home and print from a single printer at either location.
It sucks when you have a desktop at work and you can see the entire corporate network. That printer sitting across the hall might be 10 hops from you on the network but the printer on some exec's desktop might be 2 hops even though he's 2 floors up.
Or maybe you're a home user who never takes your laptop anywhere but you have a document printer and a photo printer. Or a kid and they have a printer for school work in their room. Then you have to check every time you print.
* Neither OS adds a printer by default so the scenario I described isn't really feasible and none of the OSes behave as I or the author of the article described.
If it's as I was thinking it would work without problem for the scenarios you described.
MacOS Sierra and El Capitan default to the "Last Printer Used". The UI is the same in both and there is no indication in the UI that this a per network preference.
Windows 10 defaults to "Let Windows manage my default printer" and has clarifying text under the option that states "Windows will set your default printer to be the one you used most recently at your current location". Which appears to be the feature the author loves about OSX.
Windows 7 sets the first installed printer as the default printer and when you install a printer it asks if you would like to default to it in the future.
Personally I prefer the Windows 7 approach over macOS and Windows 10. Having a network/location default would be ideal in my mind and that doesn't appear to be an option for macOS or Windows.
That's too bad.
The crux of the author's point appears to be that reducing the choices the user has to make is generally good, but I feel that modern designers are often over zealous in that pursuit.
Intelligent defaults is something that feels like a good blend. You're working with the computer, it recommends something it thinks is smart (I set this local printer as the default) and if you disagree you can modify it. One must be careful to not overly increase the mental overhead of the user this way, but smart defaults can prevent the complexity of the user trying to reverse engineer and fight unspoken rules.
This is not an explicit option, such as a default setting for printer for a named network. What I noticed back then was that printers located on the unreacheable networks were never the default offering, provided the routing could be used to distinguish where I was. So, if both home and work networks were using 10.0.0/24, it would not work.
It may be possible that binding to the name of access points is a more clever way to do it today.
I suspect because it's not really a UI problem. UI can't magically make all complex interactions easy. Fixing the problem would mean changing the requirements first... then the UI.
Edit: Never mind...missed the key sentence there in all the discussion about the 4 labels.
The point is software. There I should not spend a fraction of a second thinking, because that does not happen twice a day. It happens every 15 minutes if we are talking about me looking at my phone. UX thinking there makes a real, tangible difference.
Without the background, it just sort of sounds like "consult your UX people for the answer, always...we're never wrong".
In the US the problem however is that recycling collection is not standardized. Each city does it differently, which gets in the way of the standardization you propose. We would need a Federal standard (or at least an agreement by the relevant industry association) to help there.
It takes a bit of culture remapping, but certainly doable.
Want good examples? Spend some time in Germany or Sweden.
Here in New York, I recently lived in a building where people would, on a weekly basis, discard pizzas, soiled napkins and rotisserie chicken carcasses in the recycling bin. This was a modern building with just two different recycling bins (one for plastic/metal, the other for paper/cardboard). It's amazing how little people care.
In my current NYC neighbourhood, there are no recycling bins at all, and people have to sort and bag their recycling and put it on the street one night of the week. Not surprisingly, almost nobody bothers. Other than myself, I've only observed bags from my neighbour next door. The US can be pretty infuriating/depressing sometimes.
The person then simply matches number on article to the number on bin and throws, dead simple.
See my comment above for detailed description of this.
The problem is when items have multiple labels, because they're not made of one thing. The simplest example is Saran wrap, where the box is "paper", and the part to cut plastic is either "aluminium" or "plastic" (there are brands that make it in paper so that it's easier, though). And both parts go in different bins. Bottles can have 3 parts: plastic cap, PET bottle, paper label. All in different bins.
Then you add rules where things need to be clean, which mean in theory, your toothpaste tube needs to be cut open, and cleaned, before putting it in the bin.
And different cities have different rules.
It needs to be dead simple: number of labels = number of bins available. Maybe number of labels can be more than number of bins by 1 or 2 at most.
Secondly, use numbering instead of icons. Numbering is dead simple to match, icons require effort to match on the box against the picture on bin.
In the case of saran wrap: the box should clearly say #2 for box and #3 for contents (saran wrap). That would resolve the case you mentioned.
As for the salan wrap, the point is that it's not a box vs contents problem. The box itself needs to be broken up (removing the cutter part) before being thrown away.
If it is easier for the data scientist to sift through and correct bad data than it is for a UX professional, it absolutely makes sense to have the data scientist do the work. But the UX doesn't get to take credit for that decision.
His follow through after the initial bad analogy is then frustratingly vague; he just hand waves at Apple and gives no concrete examples of solving UX problems.
The author does provide a UX solution, which is to cut the number of bins to two or one.
Always use the word impossible with the greatest care. Everything is always impossible right up to the moment it is done.
1. Person walks up with trash in hand to 3 bins. Each bin has a poster with pictures of items which are meant for that bin. Each person has to make O(n) comparisons for each of the bin and then make the closest judgment call. Too exhausting, error prone and inefficient as each person is solving the classification problem for a given trash item over and over again.
Here's a simple solution to the problem above:
1. Setup a regulation that each article be simply labelled with a numeral marking: "1", "2" or "3". Standardize each trash bin to be called bin #1, #2 or #3. Each person simply throws the trash in the matching bin.
This moves the classification problem from the person to the time of manufacturing of the article. Also, the classification is only done once by experts and not by each person at the time of throwing away the article. Also, reduces the classification problem for the person from O(n) to O(1).
Simple solution which solves the problem in elegant fashion.
You also open the system up to folklore about what the numbers mean and what should go in each bin. And that folklore is impossible to control and correct.
It's fine if you want to build systems that way. They may be good and work well enough. But a more intuitive system will prevail at some point.
This system could be adaptable as well. So suppose, you have only 2 bins, say bins: #1 and #2. The article to be thrown is marked as #3. So, in the absence of the relevant bin, throw the article to the closest matching bin, in this case: #2. The system would be composed of simple rules that people can remember in obvious fashion.
> You also open the system up to folklore about what the numbers mean and what should go in each bin. And that folklore is impossible to control and correct.
I don't know what "folklore" means here. This efficiency of the system comes from standardization. It seems no different than rules at a traffic light intersection. One could argue that color of traffic lights could open up the system to folklore then?
Of course there are corner cases to be handled. But what percentage of the trash is Vietnamese cookie box and Japanese candy wrappers? In these cases, the person uses their best judgment.
What I'm struggling wrapping my head around is, coming up with user interfaces/experience for big, specialized applications whose value proposition IS to give the user full control over the "machine". Think GarageBand vs Pro Tools, SketchUp vs Maya, iPhoto vs Photoshop.
for example within an digital audio workstation I think it is great for quick experimentation and beginners to have some guitar pedal style plugins with just on/off switches and few knobs, that can change multiple parameters at once in a predefined way. But if you don't also give me an alternative way to control all the parameter myself the application becomes useless to me except in a few use cases.
Does anyone have recommended readings for ux projects where you can't "just" turn 4 into 1 and be done with it? (Not meant to be demeaning to the author, it is just not always possible to just remove things:))
Trash UX is really awful. In my office it's even worse - the custodial staff just dump the at-desks blue recycling bins into the same large trash can as the black waste bins. Why the heck do we even have recycling bins if it's just going to be commingled anyway?
Incidentally, is mixed-waste processing still picking up steam anywhere?
[1] http://www.waste360.com/mrfs/10-points-explain-mixed-waste-p...
There is no simple and elegant solution to knowing what goes where, since it depends on the recycling process etc.
Recycling bins at community recycling centers here just have alphabetic lists that list every item, and helpful staff that tells you were to put things if you are unsure.
People throw away the same things over and over again. Teach them once, and they‘ll throw it in the right bin.
No need to come up with a confusing icon or throw everything in a single bin.
Could that be a blog UX fail? If someone doesn't even know what you are on about from the title, then your message may not get through. To be fair, the intended audience for this might allow the author assume a certain subset of knowledge but it wouldn't hurt to cover a wider casual readership.
My thoughts on the transparent doors are that 60-70% of people will get this trash system correct. The contents made visible will promote the correct behaviour and may reduce the question of what goes where.
Thanks! I shall add it to the article.
What? You can't just say 'machine learning' about every (perceived) problem you hear.
It might be a good idea to separate paper/dry stuff from wet stuff and glass from the start, though. Just the way it's done in many countries.
(I heard that in one of the poorer regions of the city I live in, the trash is thrown into a single "bin" after collection, and possibly separated from there. Don't know if that's true though).
Applying ML to non-classification problems is when it is NOT appropriate, but this is a good pick.
Classifying by physical properties is even easier by simply classifying by physical properties. Want to pull all ferrous metals out of a pile of stuff? Just use a magnet. Want to sort tin from plastic? Blower and conveyor bins by flight distance.
Incidentally, those elements not easily covered by this kind of sorting (which has been used for many years) is also not easy to cover by magical do-all machine learning, which will have a hard time sorting plastic-coated paper from paper from plastics, just like the processes above.
ML is useful when you cannot make a clear difference using physical properties only. For example, if you treat different types of plastic garbage, you will probably want to separate bottles from other types of plastics, and you can do that with visual sorting and pattern matching.
> which will have a hard time sorting plastic-coated paper from paper from plastics, just like the processes above.
Certainly, but then this is when you still need some manual work at some point of the chain. But physical filtering + ML could reduce the human part of the work to exceptions.
(0) It isn't hard.
(1) If the trash bins are aligned in a predictable manner, then you can even do it without thinking.
https://blogs.nvidia.com/blog/2016/06/10/three-startups-find...
It's called mixed-waste processing: http://greenblue.org/reloop-what-is-mixed-waste-processing-o...
Not mixing trash is definitely easier.
If it would send it to separate process of separating the bag from its contents then I will up the game. How about a can with something in it. Or a cardboard box with bags in it.
I think you can't just slap ML on the problem and say it's done.
The best UX is happiness, beauty, convenience, pleasure, security, justice, idealism, power and so on. Did you really mean to say the best UX is no UX? :)
Dangerous. The hard part is knowing what you can take away and providing the right default for all other things.
Trying to teach every trash bin user how to sort is very difficult. It might be much easier to instead teach a few specialized people how to sort (the single stream recyclable service). This is easier for the trash users, and probably a lot more efficient as the number of sorting errors might be greatly reduced.
The food vs other items problem is probably why they explicitly said it would likely be two bins, and one only if it could be made to work.
This is the point of the article: UX is not UI. The solution encompasses changes to the entire system, not just the UI.