« Markham Panels| Main | My week »
Do we really need HTML?
I've been playing around with some radical minimal design ideas lately, and it occurs to me that I'm not so sure that we need HTML, at least for "normal" work.
Think about what most professionals do all day. You're an accountant. You're an insurance agent. You're a policeman. You're a manager. You're a factory worker.
What do you need HTML for? As an accountant, say reviewing a ledger, you need to look a couple of lists of things and compare them. The idea of a displayed list certainly predates HTML. As an insurance agent, you need somebody to complete a claim. Modern voice systems can certainly interview the person, make a stab at word recognition, and you can easily outsource the quality control overseas at pennies per form. Bingo presto, not only do you have your form completed, the system is beginning to learn the spoken word of the customer, and it was a natural question-and-answer format instead of plugging through some long form on-screen.
In fact, I've spent the last day trying to think up some combination of activities that can't be done with a blinking light, a pushbutton, and a list/text/image display.
I can't, and that's strange.
I have no idea what hyperlinks accomplish, for instance. I mean, I understand the point. And they are awesome creatures of the internet. But who's job requires a hyperlink? Sure, things need to be linked together, but the physical construct of an underlined piece of text that, when clicked, takes you somewhere else? Nobody has a need in their normal job to go to random places depending on which text is annotated. Most jobs are not that free-form.
It sounds nutty, I know. And pay attention: I am not saying that HTML shouldn't exist, or that things shouldn't be linked together, or that the semantic web isn't a good thing. What I'm asking is: simply because something makes sense in hypertext markup, should it appear on a screen somewhere?
I'm a policeman accessing a list of recent crimes. Here's the dirty little secret from systems design -- there are only a couple dozen activities for any one job that take up 99% of your time. Sometimes there is only one or two activities that you do all day long. So I push the "recent crimes" button, perhaps a physical, real button, and there's a list of recent crimes. No links, no tables, no bold text, no flashing or jumping bunnies. No fonts. Just a list. Of stuff I need to know. mirabile dictu
How far we've come.
HTML was started with the idea that the display of information would be a completely different problem than the structure of it. But look what's happened since then: it's all about display. How big the screen size is in pixels, what kinds of fonts you have (or can install), whether or not you support flash, etc.
All of this is great from a one-system-must-conquer-the-world standpoint, but completely wrong-headed from a I-need-to-separate-different-parts-of-my-life department. We're trying to invent sort of a universal generic display language. Wonderful concept, but that's not where we started out going. Or if it was, I missed it. I thought the display aspect was secondary, not primary.
But then came advertising. And money. Lots of it.
My thesis is that at some point in the last 10-15 years, the HTML web has crossed the line from being an information structure and became an entertainment medium. That's cool, and I wouldn't want to take away the goodness of the net for anything. I'm simply asking if using an entertainment medium to do your job is such a good idea. For most of us, I don't think so.
Of course, there will always be a place for the arts -- painting, writing, movies, games, music, etc -- and HTML and computers are wonderful tools for the creation and enjoyment of the arts. Most of our jobs though, however sadly, are not art.
Advertising has created wonderful general-purpose devices that can switch from balancing a checkbook to flying a F-15 in a split second. One browser page can have your investment information and the next one Facebook. And world opinion. And lolcats. Strategic investment advice. And porn. It's all one of the same. Sometimes barriers are good things. A site that pulls you in with messages from your friends can then get you to click one link -- it only takes one click -- and you've lost 30 hours playing Farmville. These are all the benefits, and drawbacks, of hyper-text markup language the way it is being used today.
And we really don't need it all that much.
How about learning what the thing is and then to start questioning it?
From the too long post it is obvious that you have very little idea what HTML is.
Have you heard about CSS at all?
So how your "just a list" would be displayed? Or is your idea, that is wouldn't? It would just float in some abstract space? How would a blind user see your list? Search bot?
[EDITOR: I was torn between deleting this and laughing at it. So instead I thought I'd leave it here. Those of you who know me understand that it needs no reply - DBM]
What we really want is magical thinking vs. Cartesian thinking. That is, thinking based on the subjective effect of the solution rather than on the objective nature of the problem.
For example: door keys, steering wheels, and volume controls have, as an objective nature, the idea of “turning.” But “turning” is not the subjective effect we are seeking when we try to unlock a door. We want the idea of “unlocking,” which is a function of key locks, combination locks, and passwords.
Magical thinking acts on the environment via these subject effects. In the Harry Potter series, for instance, there is the spell “Alohomora!” which unlocks things. It doesn’t turn things, it *unlocks* them.
The sad part is, for the first time in human history, machine interfaces have given us a real means to implement magical thinking in our daily lives. Unfortunately, as several of your recent posts point out, we go to the other extreme, and make it all about the process rather than the effect.
Neil you are spot on.
We have easily reached the point where we can computationally cover the transition from magical thinking to Cartesian thinking. But instead of working on translating tools, we seem to be adding more and more detail about the physicality -- stuff not important at all.
Put another way, what is the simplest interface? Is it pen and paper? Not at all. It's the computer asking me a question and me thinking an answer. The computer works out how to make it all happen. The next-simplest interface is probably a verbal question to which I push a button.
If we're going to implement neural interfaces, we're going to have to think long and hard about which types of stimulus we want in our brain and which types we don't. Simply because we can add detail doesn't mean that we should. Hence the topic of today's post.
Great comment. Thanks.
That's actually pretty brilliant. Wish I had something to add besides, Thanks for the fascinating read.
Excellent observation.
How very Steve Jobs of you.
Unfortunately for the shmucks like us that would prefer simpler interfaces, simpler interfaces don't give way to content. And content is king.
It's an interesting thought experiment, but I think you're undervaluing the practical utility of information design.
As an example, take a look at software development, the sort of knowledge work you're most familiar with. We focus a lot on the organization and display of information in a purely textual medium. The compiler will take anything, so we do that purely for reasons of human access and comprehension.
We stick with plain text for reasons of simplicity and tradition (even if most of us use editors that render our plain text in fancier ways to aid understanding), but that's not a necessary constraint for others. So most people use the rich set of design tools that we've mainly borrowed from the print world, which has developed them over centuries.
Those design tools aren't mere fluff. If you look at the studies by people like Jakob Nielsen, it's easily demonstrated that the sorts of differences in font, layout, and rendering enabled by HTML and CSS have very practical benefit in terms of increased reading speed, access times, and reader comprehension.
Those tools are definitely useful in task-oriented activities, because they let us convey more information and shape behaviors more effectively than we could with plain text. But I think they're even more valuable in the much fuzzier activities of communication and knowledge work, and increasing part of what people spend their time on.
Given that, I'm skeptical that you have something here. However, discussing it won't make much of a difference. If you'd like to prove you've got something, set up some wizard-of-oz tests of your interface, and just simulate all the computer activity through human activity. My guess is that you'll discover that a) you will in practice need a richer interface vocabulary, and b) developing the tech needed to make your interface effective is equivalent to solving the Strong AI problem, which will obsolete most of the tasks you're designing for. But the only way to find out is to try!
Good points, all.
And I think you've just foreshadowed an upcoming blog post :)
Yeah, I think of them as tools to make expression more pronounced. Beyond the vanilla, they allow you creative choices that plain text can not convey.