/devel

As noted in my previous post, we changed the way image navigation works on Marela. I already described “philosophical” reason for previous behavior and in this post I intend to visit other reasons and lessons learned.

I liked the old behavior because it made every page recognizable and there were no side effects, no special cases that would require different handling and make interface bigger. It made under-developed mathematician in me feel all fuzzy and warm.

What I mean by recognizable is that you could quickly tell where on Marela you are just by looking at the displayed page. Hence it should be easy to know how to proceed and what you can do. Very predictable.

Except it didn’t matter and nobody cared.

New behavior brought special cases and I’m quite certain we haven’t met all of them yet. Example of one would be navigating through set of images grouped by a tag and deleting that tag from the current image. Where does that put you?

The picture obviously can’t be a part of that set anymore so its neighbors can’t be from the set either. That’s why you fall back to photo-stream of the author. Except that most people still expect to be inside of the tag-stream. There wouldn’t be much change if we stayed there, only side-effects would be different. We could only select how to confuse users, not if.

As it happens, it’s not a big deal either. It’s confusing at first, but not for very long, so nobody cares.

What I relearned is that a tight mathematical model without side-effects, is not necessarily intuitive to human mind. In a way, we compensate for any small idiosyncrasies in a story our brains write from what we do. As with any good writing, we have guidelines that make us write better stories, but at the end it’s not they which matter, but the story itself.

Interface should only help telling it.

We recently bowed to pressure from our members and changed the way image navigation works. For those familiar with Flickr, let’s say you’ve searched for images with a given tag and selected an image from results page. When you view it, you get thumbnails of its neighbors from author’s photo-stream and not from results one. That’s how Marela used to work, but doesn’t anymore. Previous and next are now context dependent.

Our change has been universally well received among our test users and it seems the only unhappy person is me. There are several reasons for this, but in this post I’ll concentrate on just one.

How does unknown new talent get discovered on the web?

If you are a creator, how do you get your work known? I believe, like Barry Diller, that there’s a limited amount of talent in the world, but unlike him I don’t think it will necessarily get discovered. I’m not afraid of post-mortem discoveries, although it sucked to be Van Gogh. I’m afraid of talent not being noticed at all.

A friend’s wife is studying old greek texts written on papyrus. I thought these things were rare and therefore have all been read, but I was wrong. As it happens, Britain has tons of this stuff and not only haven’t they read all of them, it’s not likely they’ll do it any time soon.

So, if 10 million pages are added to web every day (and with rise of blogs I’d say this number is probably low), how big will this task be in the future? Even worse, if nobody knows of it, who’ll be diligent enough to store and preserve it?

You’re probably reading this words on my blog and there’s a good chance you came here because Google offered this link as a match to your search request. There’s a significantly smaller chance you came from one of the aggregators that follow blog’s feed or even more unlikely, you’re a regular reader who at some point belonged to first or second group of visitors.

In effect, I’ve been discovered by you (until I get forgotten), but there were several factors working in my favor. First, I write on topics which these days are fairly fashionable in certain circles. Second, because I mostly write about technological issues, my posts often feature keywords that people might reasonably use while searching and which are related to each other. It’s therefore possible to extract an idea of what I’m writing about even with rather crude tools available now and for Google to rank my page highly.

What if I didn’t? If I was a writer of short stories?

With little or no related keywords and texts without a common theme, it would be very difficult for Googles of this world to understand my writings and even less how valuable it might be to someone else. Algorithms to extract the theme of a text exist, but we are practically completely naked when it comes to evaluating quality and talent. So pages like that will never crawl out of the depths of Google index.

Since computers can’t do it, we usually rely on humans. If you like this, then you might like something liked by others, who also liked this. The problem however remains. Such scheme can only promote works that somebody else already liked and to be liked, it has to be seen first.

There’s another, more fundamental reason, why search engines and searching can’t solve this. Motivations are wrong. Unlike browsing you know what you’re looking for and need to have at least a vague idea how to describe it to be successful. That description and intent alone limits what you can find. The other problem is that search engines don’t care about variety, they care about best fit. Neither do we, users, at least most of the time.

If we want to discover, we first have to be willing to get lost.

Most people plan their vacations and so do we. There’s nothing wrong with this and I’m sure we’d enjoy them less if we didn’t. However, some of my most vivid and precious memories are of things we didn’t plan at all, which just happened against our plan.

This is what I was trying to do. I tried to create a small alley, where you could end up by accident and discover something completely new, but from which it would be very easy to get to the main street and also very obvious. It even worked, but our members didn’t like the experience.

I have no illusion that my dropped feature alone wouldn’t solve this problem, but I can’t help feeling we might have made a step in wrong direction. Maybe I tried to solve a problem which doesn’t exist, but isn’t this akin to saying we have enough culture and shouldn’t produce more?

When I was in Stockholm, I listened to a podcast of Bran Ferren’s talk at Web 2.0 conference. He made a case that knowledge workers need tools and user interfaces that put increased productivity as a goal above ease of use or learn. It’s well worth listening to, if you haven’t heard it yet.

The point is certainly valid even if mouse and other stuff Doug Engelbart invented were created with this same goal in mind. I disagree partially with iPod example, since I believe easy of use is what made iPod wildly popular instead of something like Creative Zen (which still can’t get real traction in the market), but there’s a need and opportunity for tools that make us more productive. I’d count Unix command line between them, which makes me more productive even if it’s not easy to use and even less to learn. There’s a problem though.

Nobody is a knowledge worker in every field.

Or as Alan Cooper said, we are perpetual intermediates who most of the time want to learn only as much as needed but not more. We can be experts only in very few things. We have neither the time nor inclination to become proficient with everything we do and do we have the tools to make other endeavors as painless as possible?

I think when talking about gadgets and computers the answer is generally no.

The problem lies not in keyboard and mouse. They are general purpose tools (like hammer) widely used mainly because computers are also general purpose tools and most of us use them as such. When specific needs become crystalized enough, we usually get input devices to match, like tablets for designers or joysticks for gamers. I’m sure more could be done and Bran seems to be the kind of person who will do it, but I don’t think it’s where computers fail us.

I think it’s the software, which should be either fun and easy to use or make us more productive, preferably both but by large does neither of those.

I’ll be the first to admit that iPods are rare and you’ll often find (as with Unix CLI) that you have to choose between making a more productive tool and making more fun one. But not being a beginner is not a permission to abuse me and most software would benefit if it dropped the pretense of being an expert tool and rather worked really hard on being easy to manipulate. As a rule of thumb, if you can’t charge at least 300$ for it, it’s not an expert tool.

So, this is what we are tying to do with Marela, it’s our mission. We have and will continue to work on making life in digital age enjoyable. It’s big and ambitious enough goal to keep us busy for a while.

And I still think mouse was just a brilliant invention.

According to Steve Jobs at Macworld Apple sold 14 million iPods in last quarter of 2005 and 32 million during the whole year. iPod remains by far the most popular portable mp3 player even though it’s neither the cheapest or the most feature laden of available options.

So what is it that makes iPod so popular and what does that tell us?

For the purpose of this post I’ll talk about the whole family of iPod products as one, since I believe that my points can be more or less equally applied to all of them.

1. Easy interface = happy users

Using iPod is very easy and when something is easy to do, it’s more likely to also be enjoyable.

iPod achieves this mainly by using familiar elements (playlists, slideshows…) and sticking to simple usage model. Even wheel is just an evolved volume knob from hi-fi equipment. Which doesn’t make it less innovative.

Exceptions to the mental model are expensive, since they significantly raise the cognitive burden and should therefore be reserved only for occasions when their benefits far outweigh their downsides.

2. Number of features is not important. Their selection is.

As already noted, iPod isn’t the player with most features. It’s not even the player with most features per buck.

It’s a common fallacy to think you need more features than competition. This doesn’t mean you shouldn’t add any. A product should certainly meet baseline expectations and have features most of its customers expect. But you should be careful that those it has significantly add to the purpose of the tool.

Don’t try to build a swiss army knife.

3. If you build it (expansion option), they will come (developers).

You can’t add every feature and you shouldn’t even if you could. However you should make it possible for other developers to build on top of your product. This way everybody wins.

Your budget can be smaller, majority is not bothered with features it doesn’t want or need, but somebody else can add them for those who do. That’s why web services matter and you should think about adding them if you haven’t yet.

4. Less is more (minimalism rules)

Less design, less features, less burden.

It’s like going on a trip. You can’t take everything you might possibly need with you. You have to decide what’s useful or important enough and make it fit within limitations imposed on you either by airlines or your own body.

5. Looks matter

Nobody wants to watch ugly stuff. If we are going to look at it, then it should be pleasant to do so.

Flashy designs might get your attention at first, but they age quickly more often than not. Light, understated design might not be so eye catchy, but it’s this inconspicuousness that makes them usable over longer periods of time.

To sum up, when I think of iPod, two words come to my mind. Easy and restrained. Are these the same two words you’d use to describe your website?

If you found this post interesting and would like to hear more, please come to my talk at Cyberpipe on 8th of March.

This is another news that I missed while I was busy elsewhere. Developers have added a ping attribute to link elements in the latest builds of Firefox.

Ping attribute allows website producers to specify URLs that will get pinged when user clicks on a link with this attribute. Meaning, it makes user tracking easier.

Argument being that tracking was already going on using redirects and that this feature will allow users to control this and makes browsing faster (by avoiding redirects).

I can see the logic, but I’m not sure I agree with developer’s conclusion. Redirects were at least visible to a user and he had a choice of continuing to use such web site or going elsewhere. Pings can be turned off by those, who know this option exists, but tracking will certainly be less transparent to everybody else.

I’m still on the fence about this one. How about you?