I like the author's honesty; he seems to hate decision-free computing but acknowledges that it's what most people prefer.
If his main objection is that (like everyone's saying) abstractions are sad because they make tinkering harder, then I'm with him. But the question is whether abstractions and tinkering are mutually exclusive. The examples I've seen seem to suggest that they are. But it would be a shame if that were true, because I think both are very important.
Spot on. And the question you have is quite similar to the question I have, but I'd phrase it differently;
Are low cognitive friction abstractions implicitly the exclusive domain of a locked down ecosystem? Or is it possible to have one's cake and eat it too?
I think it's very clear that it's much easier in a locked down ecosystem, I'm just not certain beyond that.
It feels to me that the problem isn't so much the the total cognitive load/friction of a user interface, but the kind of thinking an interface demands. In order for an interface composed of high-level abstractions to be something that we can tinker with, it has to be a system that follows logical rules that can be easily inferred, and those rules have to be Turing-complete. Once you have formed an accurate mental model of the software, you can use it very efficiently with a low cognitive load.
But ordinary, everyday people don't want to infer how a system works. They don't want to think scientifically, and they won't invest any effort in building a mental model of how the device really works. Instead, they want the device to conform to the mental model they cobble together on the fly - logical consistency be damned. They have no curiosity to learn how to fully use the tool (or else what little curiosity they have is overwhelmed by the frustrations of the software violating their expectations). If they did, they would be programmers, because anybody who understands how computers work will end up using that knowledge to make it do something new. (The power and possibilities of a programmable machine are irresistible to anybody who actually comprehends it, but first you have to overcome the initial task of understanding how the machine computes. And to continue as a programmer, you have to tolerate the frustrations of the leaky abstractions.)
If you want a user interface that can be used without any scientific thinking, you have to restrict it to a finite set of capabilities so that you can anticipate what the user will try to do. But at that point it is no longer really a computer: it's a shiny toolbox. You can't make new tools with it; you have to buy them at the App Store.
The problem of users being unwilling to think analytically isn't one that Apple can overcome. I doubt that any for-profit corporation can do anything about it, because it is a broad cultural and generational problem. Computers haven't been around long enough for us to know how well the general population will eventually adapt to using them, but it's a sure thing that without true AI, people have to adapt their ways of thinking in order to use the full power of computers. Until then, people will keep wondering why we programmers can't manage to give their Photoshop an "enhance" button that works like the ones they see on TV.
I agree especially with the aspect of once you grasp the underlying truth of what the thing really is, you can't resist embracing it and making it a part of you. It almost sounds like a choice between ignorance and enlightenment, but history tells how rare the latter and how common the former. Perhaps the entire endeavour is simply not conducive to normal human psychology.
If we are to assume that this is the case, is the ultimate result something of a guarantee of a class of professional developers marketing their wares to an ADHD afflicted mass of semi sapient, sycophantic, glad-handling humanity?
I want another path, I know wanting something doesn't make it real, but I can't help it on this one I don't think.
There is solace in the fact that we can build really powerful tools like web browsers with social networking apps or CAD programs and normal people can use them if we design them well enough. But Turing-completeness hasn't gone mainstream, and it doesn't look like it will anytime soon. So that leaves us hoping for a breakthrough in AI, because that's what it will take for a computer to be able to always understand the user's intentions.
> But ordinary, everyday people don't want to infer how a system works. They don't want to think scientifically, and they won't invest any effort in building a mental model of how the device really works.
That's not true ... "everyday people" (whatever that means) always have a mental model of how something works (be it a computer, a car or a washing machine) ... that's how humans think, the problem is that their mental model is usually different from reality.
And you can almost always put blame for this on the interface or on the education given to them.
It's like you're saying I should know how the internals of my car works (lots of science in there too). Well I don't ... all I know is how to drive it, and I've had problems because of that ... but it's not like I care. If it ends up bugging me so much, I'll search for a smarter car.
Regular users are lazy indeed ... some of them just don't get it, others are doing just fine for their day-to-day needs. Those that are doing fine, like my wife, tend to have a good teacher nearby ... so the only meaningful way of countering their laziness is through proper education.
I know that people always have a mental model of how something works. But it's rare that somebody actually stops to consider whether their mental model is accurate or logically consistent, and unless they do, it is neither. If your mental model of a piece of software is self-contradicting, then it obviously can't reliably tell you how to use the software.
There's a lot you can do in designing a user interface to try to make sure that the user develops an accurate mental model, but it gets harder as the software gets more complex and more powerful. A car's controls are a pretty good abstraction of it's drivetrain, but most of the complexity of a car exists for the sake of efficiency and increased power, and it's a physical, observable system anyways.
I think that once a piece of software becomes Turing-complete, it is impossible for user interface affordances to lead a lazy user toward an accurate mental model. After all, to program, you have to be developing a mental model of what you're creating, within the context of the programming environment. That's two mental models that both have to be accurate, and one of those can't be anticipated at all by the user interface designer.
Programming is inherently harder than almost any other thing we do with computers. If a piece of software is Turing complete (and thus subject to tinkering), then the mindlessly-easy-to-use subset of it's capabilities can't be Turing complete.
8 comments
[ 3.6 ms ] story [ 19.8 ms ] threadIf his main objection is that (like everyone's saying) abstractions are sad because they make tinkering harder, then I'm with him. But the question is whether abstractions and tinkering are mutually exclusive. The examples I've seen seem to suggest that they are. But it would be a shame if that were true, because I think both are very important.
Are low cognitive friction abstractions implicitly the exclusive domain of a locked down ecosystem? Or is it possible to have one's cake and eat it too?
I think it's very clear that it's much easier in a locked down ecosystem, I'm just not certain beyond that.
But ordinary, everyday people don't want to infer how a system works. They don't want to think scientifically, and they won't invest any effort in building a mental model of how the device really works. Instead, they want the device to conform to the mental model they cobble together on the fly - logical consistency be damned. They have no curiosity to learn how to fully use the tool (or else what little curiosity they have is overwhelmed by the frustrations of the software violating their expectations). If they did, they would be programmers, because anybody who understands how computers work will end up using that knowledge to make it do something new. (The power and possibilities of a programmable machine are irresistible to anybody who actually comprehends it, but first you have to overcome the initial task of understanding how the machine computes. And to continue as a programmer, you have to tolerate the frustrations of the leaky abstractions.)
If you want a user interface that can be used without any scientific thinking, you have to restrict it to a finite set of capabilities so that you can anticipate what the user will try to do. But at that point it is no longer really a computer: it's a shiny toolbox. You can't make new tools with it; you have to buy them at the App Store.
The problem of users being unwilling to think analytically isn't one that Apple can overcome. I doubt that any for-profit corporation can do anything about it, because it is a broad cultural and generational problem. Computers haven't been around long enough for us to know how well the general population will eventually adapt to using them, but it's a sure thing that without true AI, people have to adapt their ways of thinking in order to use the full power of computers. Until then, people will keep wondering why we programmers can't manage to give their Photoshop an "enhance" button that works like the ones they see on TV.
If we are to assume that this is the case, is the ultimate result something of a guarantee of a class of professional developers marketing their wares to an ADHD afflicted mass of semi sapient, sycophantic, glad-handling humanity?
I want another path, I know wanting something doesn't make it real, but I can't help it on this one I don't think.
That's not true ... "everyday people" (whatever that means) always have a mental model of how something works (be it a computer, a car or a washing machine) ... that's how humans think, the problem is that their mental model is usually different from reality.
And you can almost always put blame for this on the interface or on the education given to them.
It's like you're saying I should know how the internals of my car works (lots of science in there too). Well I don't ... all I know is how to drive it, and I've had problems because of that ... but it's not like I care. If it ends up bugging me so much, I'll search for a smarter car.
Regular users are lazy indeed ... some of them just don't get it, others are doing just fine for their day-to-day needs. Those that are doing fine, like my wife, tend to have a good teacher nearby ... so the only meaningful way of countering their laziness is through proper education.
There's a lot you can do in designing a user interface to try to make sure that the user develops an accurate mental model, but it gets harder as the software gets more complex and more powerful. A car's controls are a pretty good abstraction of it's drivetrain, but most of the complexity of a car exists for the sake of efficiency and increased power, and it's a physical, observable system anyways.
I think that once a piece of software becomes Turing-complete, it is impossible for user interface affordances to lead a lazy user toward an accurate mental model. After all, to program, you have to be developing a mental model of what you're creating, within the context of the programming environment. That's two mental models that both have to be accurate, and one of those can't be anticipated at all by the user interface designer.
Programming is inherently harder than almost any other thing we do with computers. If a piece of software is Turing complete (and thus subject to tinkering), then the mindlessly-easy-to-use subset of it's capabilities can't be Turing complete.
Recently a non-geek friend of mine bought an iPhone. She already had if for over a week when I met her and finally showed her how to zoom...
Not saying the iPhone interface isn't great, but it is still not problem free.
Also, doing specific things might require using iTunes (to buy the required apps), and the iTunes interface is definitely not great.