As I constantly iterate on the design of Fury in my head, I'm influenced here and there by things I read or anecdotal experiences I have. Today's post by Phiipp Lenssen, Context, not Navigation, is having a big impact on the virtual-Fury in my head.

Most importantly, it resonates with my awareness that the experience and motivations of the everyday reader are completely different than the google visitor, and the look and feel should reflect that.

Categories were all the rage, and are de rigeur for most blogs nowadays, but they don't scale well at all. They tend to work best when the branching factor is constant, that is when there are roughly as many items in a category as there are are categories in total. Another way of putting it is, if each post is only in one category, then your number of categories should be roughly sqrt(number of posts). This doesn't scale well when you reach 2000 posts and 45 categories, with 45 posts in each category. I actually have 91 categories, because I'm inefficient, and because many posts are in multiple categories, and, well, I am a freak.

Anyhow, the article's very thought-provoking, and I'll have to see how it impacts my twin desires to further granualize and consolidate Fury's organizational structure. I should talk more about this soon. Maybe I'll even have a demo.

Comments? (8)

It's tempting when designing a page to just design 'above the fold', that is, the things that the user sees without scrolling. The term comes from the newspaper industry, where half of the front page is 'above the fold' and the less important half is 'below the fold'.

It's interesting because in newspapers it's a 50/50 split. In tri-fold letters it's a 33/33/33 split. On web pages though, especially weblogs, the majority of content usually exists below the fold.

Sippey gives a great viewpoint of exactly what several popular weblogs look like if 'the fold' didn't exist. It's got me thinking about how the value and function of sidebar navigations changes as one descends into the depths of a page.

Scott McCloud (of Understanding Comics fame) uses this perspective extremely well in his online comics, starting from the beginning.

My mind boggles at the possibility of melding Scott's comic model with the inverse chronology of a weblog...

Comments? (11)

I try not to post what everyone else posts, but when there's such a confluence of memes, it's hard to resist.

According to Edward Tufte, PowerPoint is evil. It helps speakers present and audiences tune out (in PowerPoint form). PowerPoint is responsible for the destruction of the Space Shuttle Columbia, and its principles can be seen in the destruction of the Challenger as well (in powerpoint format, ironically). In short, PowerPoint makes you dumb.

What if Lincoln had access to PowerPoint at Gettysburg? (by Peter Norvig who, in addition to co-authoring my undergrad artificial intelligence textbook, is also a fellow Googler)

David Byrne, however, has found a saving grace in demonstrating that PowerPoint can be an artistic medium. If anyone has a link to the actual content, I'd love to see it.

Comments? (9)

The New York Times did an impressive flash presentation of geographic variables in today's recall vote. I particularly like the distribution chart of how often a given candidate's name appears on the first page of the ballot.

Comments? (3)

Sorry for the slowness of posts. This time though it's completely warranted. I've been putting a lot of effort into Fury 4.0. Right now most of the obvious work has been in the visual design of the front page.

That visual design incorporates some changes in the interactive design of fury, but most of the interactive changes (and they are substantial) will be slower in coming.

I feel the flows have to be validated more, and will require more user testing, while I feel qualified to take a good solid stab at the visual redesign based on my knowledge of user habits, relative importance of various elements, and personal asthetic.

I'll probably have the first static visual design mock up late tomorrow or Thursday. Before that I expect I'll dive in to explaining one or more of the redesigned flows. FYI, I'm redesigning (or designing the initial flow) for: reading comments, posting main comments, posting IMblog comments, registering/login.

And that's just the stuff I can recall off the top of my head.

I'll say this though, after working on the VisDe for 4.0 for the last two days, it's almost icky goving back to 3.2 to actually post...

Comments? (2)

So hey, I've long found it amusing that so many webloggers with such tight design skills do their redesigns in private, suddenly unveiling them to the world with a big "here it is!"

Trouble is, this is the antithesis of the traditional interaction design process. Showing the design around to a few of your friends shouldn't be a substitute for an actual usability experiment, for a lot of reasons, most notably an inherent bias towards the designer, a familiarity with the existing site (this is useful, but naive users should also be tested), and most importantly, the fact that a person's subjective opinion is not the same thing as the usability of a site.

So I'm going to (more or less) conduct the Fury 4.0 redesign by the numbers. I'm going to do some low-fi prototype testing, some task analysis, a smattering of cognitive walkthroughs for the identified common tasks, and some rolling usability testing as the redesign comes along.

I've already started with a lot of logfile analysis of Fury 3.2. I've identified the five categories of visitors, and their use patterns:

1. The regular subscriber - You read this site at least once every two weeks, and get here either via a bookmark, by hand-typing the URL, from your RSS aggregator (desktop app, or web-based aggregator), or by a link on your own personal links page (I'm so tempted to link to some of these as examples, but they might be private, so I won't).
   
   You read comments. Most of you use the timeline bar at the top of the screen. You might lurk, or you might post. Most of you check in at least 3 times a week. Some check several times a day.
   
   You almost never visit anywhere other than the front page, unless I linked to it in a new article.
    
2. The general referred user - You saw a link on the sidebar of another site and decided to check it out. You might look around a bit. There's a 30% chance that you'll click on one of the topics in the 'Read by Topic'. If you do, there's an 80% chance that you'll follow the 'sex' topic. There's a 90% chance that you'll be disappointed by it.
    
3. The specific referred user - A blog or news site you read linked to a specific article on fury, and you followed the link to check it out. When you do, you might visit the front page, and it's just as likely that you'll click on the 'Bio' link to find out more about where you are. You're more likely to become a regular reader than any other group.
    
4. The google searcher - You dive in and dive out. You'll almost never go anyplace other than the page you land on (unless it's to the aforementioned 'sex' link). On rare occasion you'll actually leave a comment, but if you do, there's (almost exactly, strangely enough) a 50% chance that you're either a crackpot, conspiracy theorist, extremely vulgar, immature, or some combination. Trouble for you is that, though you don't know it, almost nobody will ever know you left that comment, unless they constantly scour the archives for recent comments.
    
5. Kevin Fox - I use Fury differently than everyone else. When it doesn't suit my needs, I've built hacks in the back end so that I can do what I want. When someone leaves a comment on the site, it automatically emails the comment to me, so I never have to check back to find new content, and can respond to comments right away.
   
   I have hidden pages where I can get up-to-the-second lists of who came to what page of the site, and where they came from. I can spot new referring links easily, and see how popular that link is.
   
   I use the timeline bar as a guilt-o-meter, always wanting to see at least some dark blue on the page, lest I feel the page is growing stale.
   

I've spent the last several months with these use patterns in my head, and they have driven a few changes over the years (the timeline, color-coding, permalinks, comments, etc.) but now I've got enough new ideas that I'm doing a complete rewrite. I closely considered making it 100% CSS, but I found that while the concept of CSS is extremely elegant, in practice the compatibility differences amongst my target browsers (even between IE 5.5 and IE 6.0, browsers of the most common Fury visitors) mean that I'd have to code many kludges just to make it work right, and it would make me more reluctant to institute changes, knowing a small change could wreck the site.

Instead I'm using tables for layout, and CSS for styles, as I mostly do now.

Okay, this post is getting much longer than I intended. Back to the interaction design model, I want your input. I'd like the regular users to be the eyes over my shoulder, I want you guys to play the role of the stakeholder. I'm designing a tool both for you and the other four groups, and while I'll take your comments for what they are, and not gospel, I realize you guys have a lot of good ideas and frustrations, and as long as this post's comments inform the design, and don't drive it, I think it'll make for a better redesign all around.

So, knowing that many of the labels in the following wireframe will need description I won't delve in to until later this week, I'd like to share the preliminary Fury 4.0 home page wireframe.

Questions? Comments? Go for it.

Comments? (38)

Trust the government to spearhead waterfall design at the expense of usability. They're trying to make governmental data easier for ordinary citizens to find, but their 'three clicks or less' mantra leaves a lot to be desired.

"Three clicks or less" sounds great in meetings and when pitching to corporate schmoes, but it has absolutely nothing to do with usability, beyond ensuring that the final product will have been crippled by a false constraint at the outset.

The joining of several databases into a few unified search databases is laudable, but search has so much to do with how search requests are understood by the system, how different results are given levels of significance, and how those levels are indicated to the user, that the most unified search engine can end up being the worst, unless these factorsare taken into account.

Case in point: Go to 4 out of 5 consumer electroncs sites and search for a product name or part number and you'll receive 23 press releases that mention the product name, and you have to drag through two or three pages of search results before actually getting to the product page. This is a particularly lamentable example because it's clear that users desire product pages over press releases, and they could easily be grouped first, or the result set could even be gathered into piles from different categories so the user can say 'ooh, press releases!' and dive into that subset of the results.

Hopefully, government info is just as structured and easily clustered. They also have the benefit of being able to enforce metadata inclusion, to allow better sorting and grouping of result sets based on meta tags.

Of course, my grousing is based on the PR-speak coming from the project, and I'm assuming that the design will follow the propaganda they're spouting to the press. I just hope the actual designers don't accept the 'three click' mandate as the backbone of design, because just because you can get anywhere in three clicks, if each of those clicks are from a palette of a hundred or a thousand, then usability was gone before click one.

Comments? (61)

I've been doing a lot of thinking over the last three months on the format of a blog. The design patterns of this type of site, and those that set it apart from a macrom site, or a static informational site.

Of course blogs are more timely, and are therefore stickier (mmm... sticky blogs...) but does this advantage come at a price?

They typical blog has a front page and date-indexed archives section. Some more sophisticated blogs also have categories, so you can find posts loosely related to the one you're reading, or look for things on a particular topic.

Having had both of these 'windows into the past' for a while, I don't think they're sufficient. I still look at some static sites and miss in my own site the qualities of relevance that they have.

There are two kinds of posts (okay, there are as many kinds of posts as there are posts, but for the sake of this post, I can make my point by dividing posts in general thusly): Those which have meaning within the running commentary of posts, or otherwise are relevant specifically to the time when they're posted, and those which, insights, information, commentary or otherwise, are items that would make it onto a static site, if that was what you kept. This second type of post is the kind of thing you wish people coming to your site for the first time could see when they're trying to get a foodhold understanding of who you are, rahter than forcing them to dig through sedimentary banality, or lurk long enough until they think they know you.

So this post isn't saying much more than that my current focus of blog framework study is looking at the more effective kinds of information presentation on static sites, personal or otherwise, that have a fair amount of data, yet easily allow people to self-select the kind of information they want. I hope to identify ways of building this kind of framework dynamically, and incorporate it into the blogging system, so that when I write what I think is a profound, timeless, or otherwise worthy piece, along with filing it dutifully away in the date archives and a few topic pages, it'll also find a home in the pantheonic 'static site'.

Clearly I'm thinking far too much about this for someone with a backlog of posts and only twelve days remaining before the Big Drive, but then there's always spare cycles to burn, walking down the stairs, showering, sleeping...

Hey, happy Monday, y'all!

Comments? (53)

I wonder what would happen if, as a design exercise, a designer were asked to take an existing site and create the 'next generation' of that site. Flat photoshop mocks a flow chart and a sitemap.

Now take those docs to another designer, tell them this design has been around for over a year and is stale, and ask them to make the 'next generation' of the site.

Repeat through several iterations. I wonder if you'd just get different design styles, or if each designer would actually innovate, using the previous designers work as a base?

Sadly, information architects and site designers all too rarely use real metrics of demonstrated usability from existing sites when trying to revamp them. I wonder if in those cases it would be better to skip the trouble of actually building the 2nd and 3rd generation sites, and go right to the 4th, 5th, or 10th?

Of course, the real answer is to use a designer who finds out what the actual problems and successes of the previous design were, but this might make an interesting exercise, ala Photoshop Tennis.

Comments? (45)

Okay, so I did it! The logic took a looot of thinking through, but I've implemented 'new to you' color coding!

Here's how it works:

• The first time you come to the home page since I installed new to you, you get issued a static cookie. Initially, you're fully up to date, and nothing is 'new to you'.
• When new things get posted, they'll show up in the red color on your next visit. Things that are color-coded this way are the message title bars, the 'timeline' bar at the top of the page, and the little 'comment' circles associated with message blocks in the timeline.
• When you visit, look at the 'timeline' to see which messages have new comments since your last visit, and which messages are new since your last visit. you can click on either the message block or the comment circle to have the associated content spring forth.

Here's the slightly tricky part:

• If you visit the site and there's say, 4 new posts, and you follow a link and come back to the site, it would be a bad thing if the system decided that this was another visit, and thus all these things should be marked as read.
• With this in mind, I coded a system where a 'new visit' is made only if the home page view happens more than X minutes after your last home page view. Basically, if you look at the home page every minute, things that were new will keep showing up as new, but when you leave for X minutes and come back, the system will say everything that was posted since the last page view (not counting the current one) are new to you.
• This all sounds horribly complicated, but the point is that you-the-reader shouldn't have to think about it at all, and it should just work like you'd expect.
• 'X' is currently set to be 10 minutes, so as long as you look at the home page once every 10 minutes, Fury will think it's part of the same 'visit'.
• If you want to reset your 'last visit' timer by hand, click on the "New To You" link in the Legend navbar, top right. that will force everything to be marked as read, and will reload the home page.

Make sense? Yes? No? Don't worry about it. If I'm any good as an interaction designer, it should all make sense without my explaining it (except for the 'mark all' easter egg), but I like to keep you posted, and I'd like to hear what you think.

Comments? (34)

Erik asked:

"What are the pitfalls you mention to doing a 'new to you' widget? interested because i've been thinking along those same lines recently."

Thanks for asking! It's two things mostly: First is about how you cookie. If you always cookie with a simple permanent 'last visited' cookie each time they come, then if they follow a link and then come back, all the rest of their 'new' stuff is gone because you 'just visited.'

This necessitates two cookies. a 'true just visited' permanent cookie, and a placeholder cookie with a 5 (or 10, or 60) minute timeout that says 'before this session, they last came at such-and-such a time' and whenever they hit the home page, if that cookie exists, it gets renewed with that timestamp, so they don't 'reset' the new until 5 (or 10 or 60) minutes after the *last time* they hit the home page.

That's cool, nifty, and doable, but then what if you use more than one computer? What if you want to be able to share that cookie across several computers? Well, then you have to sign in in some fashion or another, and then Fury has to keep track of modtimes, so it can share them between computers. Without this it gets annoying to see things as 'new' when they aren't really, just new to that computer, and ends up being a detractor because it's giving misleading information.

That's what makes it so complicated. :-)

Comments? (32)

So I made a few tweaks to the timeline code. The links are now 'normal' and the title of the post is in the 'title' tag of the link so most browsers will bring up the title as a tooptip if you hover over the box.

I also moved the '?' to the left and made it 'Recently:' which should do a better job of giving the timeline a context, and hint to its meaning to the new user.

I'm probably done futzing with it for the moment. I may implement a 'new to you' piece of functionality, but that has its own pitfalls. For now I'd like to let it sit. I'll be asking in a week or two whether you actually use it. :-)

Fickle: Okay, I nixed the 'recently' bit because having any text at all made spacing inconsistant across browsers. That's the aesthetic reason, but the real reason is that I like how it looks without indication. I think there's something fascinating about an interface that you get to learn. Note that you don't have to 'get it', as it's perfectly usable without ever knowing that the bar does anything at all. There's an 'ooh that's neat!' quality to discovering things on your own.

Call it an easter egg.

Comments? (31)

So the observant reader might be wondering what's up with that bar of boxes just beneath the masthead. (err, on the main page, not if you're looking through individual articles)

It's a new way to navigate and visualize the page! In a nutshell, each rectangle is one story on this page. The first (leftmost) one symbolizes the top story on the page. The color of the box is the same as the header color, which indicates whether the story is less than 24 hours old, between 24 and 48 hours old, or older than 48 hours.

Clicking on the bar will jump you to that story in the page. This is intended to make it easier for people who come once a day. They can just click on the last dark-blue bar and jump to that article, and can scroll their way up after that.

Similarly, and possibly even more useful, the little circles at the right end of the boxes indicate whether that article has any user comments, and if so, how old the most recent user comment is. that way you don't have to remember how many comments there were or, worse yet, keep checking comments to see if there are any new ones. Just look for the blue circles for recent comments. Depending on how late I stay up tonight, clicking right in the little circle may also pop up the comments window directly.

I'd love to hear what you think of the functionality and implementation, and whether you'll use it in practice.

Happy Monday!

Response to feedback: Making it more intuitive - One of my earlier designs called for summing up each post into a word and putting that single word in the post's box. I decided against it because it's too visually distracting (and because I dislike changing the database schema every time I make an interface tweak) but I might give it a try. I could have it in grey, turning to black on the rollover. This would also provice a convenient solution to issue #2:

Anchor tags aren't working in Netscape/Opera/Lynx - Yeah, they use the onclick property of the TD tag, and a lot of browsers can't deal. I'd use a transparent gif with an A tag, but I can't make it stretch correctly. As long as the comment box is sometimes there, stretching it to a percentage will either result in nonclicable box areas or comment boxes that wrap to the next line, both of which are worse. The other problem is that there's no affordance of clickability, since the cursor only changes to the finger when you're over the comments button. Lastly, clicking on comments brings up the comments and drops you to the post, which I'll cop out and say is a feature, not a bug, but if you're just catching up on comments, you probably don't want to scroll to the top every time you want to check out another post's new comments.

It won't make sense after this post falls from the top - I just added a "?" at the end, which links to this post. Let's all just smile and ignore for a second that the bar only appears on the home page, and the explanatory post comes up on a page that doesn't even have the bar it's trying to explain. I have an answer for that, but it's part of the Fury 4.0 redesign (oooh Fury four-point-oh... Okay, so it's just me), so it'll have to wait.

Length mapping makes some boxes too short - Fixed. I added a 'minimum size' buffer, so the really short posts don't look as short. This is actually more accurate because originally it used the number of bytes in the body to compute the %, but it didn't factor in the real estate of the title and the box, comment link, and all that. Now things won't get super crowded unless you have your window pretty narrow and I make a lot of posts in a week.

One more question: Does this bar just make sense for the home page, or would it be useful on the topic and date pages as well?

Comments? (20)

So one of my own personal complaints about Fury? There are 110 links on the front page. If you live in Southern California, you might understand when I say that I feel like my front page looks like the menu at Jerry's Deli. If you don't, you'll just have to take a look at the web version of Jerry's menu and realize that the physical version is at least four times worse.

I may go more hierarchical, or allow people to choose the modules they want to see, ala My Yahoo, or I may find some other completely novel solution which, by definition, will have to be learned by the user and is therefore bad.

Comments? (98)

Watching ST: Voyager I was thinking about 'Voyager's database' and the sci-fi assumption that everything in a computer system is in a 'database.' I suppose that broadening the term to simply refer to a repository of information would mean that any hard drive, isolinear chip, or filing cabinet is a 'database of information' but then I starting wondering what it would look like if all data really were in a database.

On one level, that would be easy to do: Simply take each file and put its ascii or binary data in a database field, and put its name, modification date, file type, and other metadata in other fields in the same entry and voila, it's in a database, albeit a shallow, general purpose one.

But what about the files themselves? What if, in an XML-like fashion, every file's data were in turn kept in a database? What if individual file contents consisted entirely of its own database, with its own schema, stored procedures, and all the rest? For example, one of the nifty things about mySQL is that each table is kept as a file on the hard drive, and you could back up by exporting a dump of SQL calls needed to reconstruct the database, or you could just copy the table file and put it in a safe place. What if these files were in turn kept in binary format in the fields of another, encompassing database? So that regardless of differences in schema from file to file, data would still be kept in databases within databases, or hierarchical databases?

To take it a step further, what if the SQL language (I hate saying that like I hate saying 'ATM machine' 'RAID array' 'PIN number' or any other redundant acronym) were extended to allow for 'deep queries' that would be able to isolate records and then further qualify and manipulate contents of the databases stored within the data of those records themselves? Excel spreadsheets could be stored in a database format, alongside powerpoint presentations, word files, email repositories (with records being individual email messages, which in turn could have attachments with embedded databases), and so on.

On some level what I'm talking about boils down to hierarchical XML, but with the power of SQL built into it. XML is an ideal format for transmitting information, as databases are for storing it. XML has the added features of DTDs to specify the format particular datatypes should follow, while SQL has unique schemas. Perhaps they both could learn a bit from each other.

Anyhow, no rousing conclusion here, other than to say I've got to get to bed. Stayed home sick from work today, went to the doctor, and the fun goes on and on. I just wanted to type this one into the ether before going to sleep and having it replaced by random dreams...

Comments? (65)

Last night I went to the monthly UI/IA cocktail hour, a gathering of information architects and people in associated positions or bents. We did a lot of talking about the iterative cycle and how in reality it's not "iterate, iterate, iterate" but "iterate, obliterate, iterate."

The problem tended to boil down to the perception of the web as a media, and thus any web project is something that grows stale and needs to be redone, just by virtue of it being media, like an ad campaign or a movie. Of course that's not the only problem, user needs, business goals, and functionality changes, so that the house that was build to fit the needs of the time becomes inadequate, and a house that is constantly being appended with additions becomes ungainly and difficult to navigate.

Basically, these projects shouldn't be seen as houses: when designing a project, we should realize that half the functionality will be obsolete in a year, and new functionality will be required, so designing a house isn't the best way to go. Instead, create one unifying layer, a design paradigm that sits above the functional layer, a look and feel that could be used for any number of purposes and modalities. Then, with that 'interactive style guide' in place, design functionality into modules, so that as some functionality is obsolete, it can be removed, and new modules can be brought into play, without leveling the house and 'doing it right this time.'

The main thing is to realize obsolescence is that building the 'perfect site' (or application) isn't about making a functional work that will endure through the ages (if your functional needs don't evolve over time, then you should be worrying about the stagnancy of your business more than your website), it's about creating something that can evolve bit by bit, instead of revolve every 18 months into something 'new and this time perfect.'

Of course, in a practical sense, this means abstracting content from display, in general abstracting front-ends from back-ends, and creating small, tightly focused mini-houses and shaping them into a neighborhood. Like a real neighborhood, some houses will get torn down and new ones may be built to replace them, but since they're small segments, there is never a jarring time when everyone has to learn something new.

In a sense, it reminds me of company earnings statements. Some companies always seem to show a quarter-by-quarter net profit, but only after you disregard certain 'one-time restructuring charges' which is fine, except that every quarter seems to have these charge, either because of layoffs, acquisitions, or other 'extenuating circumstances.' Similarly, if you're spending most of your development cycle waiting for those 6 months when the system perfectly maps to user and company needs (of course, those are the same 6 months when people aren't making full use of the system because they're still learning how to take best advantage of it) then you're riding a jump-and-stagnate bleeding-edge curve that keeps IAs happy, developing new metaphors, architectures, and design and evaluation processes, but at the expense of an actual usable and perpetually adaptive product. Better to drive a car and keep it on course by making minor corrections every second than to point it in a direction, let it go for a while, then see where it is when it runs out of gas, fill it up and point it again, and repeat.

Okay, enough with the metaphors, I'm late for work!

Comments? (12)

Since Benjy has been playing around with the back-end templating system I wrote for Fury 3.1, I've started giving names to some of the terms and constructs I've used. One of the most important is the concept of 'pseudostatic pages'. Basically, every permalink (http://fury.com/article/*.php) and the topic pages (http://fury.com/topics/*.php) all appear to be static URLs. They don't have parameters tacked on to the end (article.php?id=784) and so search engines will actually index them, while they don't tend to index parameterize pages.

In reality, these article and topic pages don't exist. The actual page for displaying individual articles is a page called 'article' and the one for topics is called 'topics' (don't ask why one is singular and the other is plural. 'articles' actually exists as well. I should probably change it to be consistent). So when you ask for http://fury.com/article/890.php it actually calls http://fury.com/article with the parameter '/890.php' which the code strips down to 890 and displays the data accordingly.

I did this partly to be indexed by search engines better (read: indexed by search engines at all) and partly to have a cleaner system, where a layperson can take a look at a url and make a fair guess at what they'll see if they go to http://fury.com/topics/interface.php.

My original design accounted for making small unique 'keyphrases' for articles as well as topics, so that this post might be found at http://fury.com/article/pseudostatic.php and I'll probably implement that soon, but I'm loath to go back through a thousand previous posts and make up keyphrases for them, so I'll make sure that the old way will still work.

Anyhow, just wanted to share a bit about the underpinnings of Fury. Benjy's giving me good ideas on ways to make furynodes (another neologism, this time referring to the whole fury template engine) more accessible to the average user, and eventually I suspect I'll be releasing a version for people to use, in similar fashion to greymatter, to create their own blogs and sites. In actuality, it's more geared towards making templated web sites with hierarchical templated objects within objects, more than a weblogging system per-se, but we'll see where it goes before release.

Comments? (40)

It stuns me that so many people develop and iterate web sites (not just personal sites, bt multimillion dollar marketing sites) in complete absence of server log data. They don't see the vitality in looking at the logs, relying instead on user testing (server logs are user testing!) or worse, personal intuition.

Your server logs are your map, and driving without them inhibits you from staying on the road. Aggregation log analysis tools like analog wusage and webtrends are great (and Personify really is great, if you can afford it), there isn't really a substitute for getting into your log files and actually watching a user's path through your site. If you do this now and then (using grep to isolate an IP and 'tail -f' to get a running realtime output stream ala the Matrix) you start to get a real feel for the audience on your site. You understand where they go, how long they think beforemaking a decision, what they like, what they don't, and what, in particular, makes them come back.

If you run a site but don't have access to your weblogs, get them. They're the holy grail of your readers, and they'll teach you things you'd never expect about how to make your site better, not just for you, but for the people who matter.

Examples forthcoming...

Comments? (1)

A while ago I talked about the need for subdomaining, that people should be willing to sell or rent subdomains of their domains to others, especially if theyhave domains that could be of use to a lot of people, like cars.com. (nissan.cars.com, ford.cars.com, etc.) Really, there should be more TLDs, but until then (or for less common names like kfox.weblogs.com or fox.clan.net) there's IHN. These folks will handle the DNS end of it for you, and let you build traffic to your own site (and make money) by leasing (or whatever) subdomains of your SLD to other people.

They have a few subdomainable domains of their own (like clan.net), and they encourage others to add their own domains for subdomaining. Worth a look see to get an idea of where the net might be going, if nothing else.

Comments? (10)

It amazes me the fervor people get over domain names. One company sues another company (or, all too often, an individual) because the second party has a domain name that the first party wants. Nissan Computers is a good example, as they are currently being sued by Nissan Motors, both of which have the trademark "Nissan" for their area of business. If the lawsuit doesn't put the domain in the hands of Nissan Motors, you can bet their next stop will be ICANN, to try to take the domain name by force.

Why does one company or individual have a greater right to a domain name than another? the US Courts recently declared that a domain name isn't property (by judging that it can't be stolen), so how can it be taken away from one party and given to another by court order? If Nissan Computers starts selling cars on their site, I could understand an injunction ordering them to stop, but not one ordering the domain to be transfered to another company. And if they don't even try to sell cars, then there should be no basis at all.

If I start my own ISP (called, for the sake of argument and lack of imagination, AOL), and I give each user a unique screen name, can Apple Computer come in and demand that I take away Joe Blogg's account name ("Apple") because it's the same as their company name?

But I seriously, seriously digress from my intended topic...

Subdomaining: TLDs are a little overrated. We are all children of the dot-com. Woe betide the lesser children of the dot-org or dot-net. It's yourcompany.com or bust. Along with that of course come the obligatory domain names www.yourcompany.com, ftp.yourcompany.com, and all too often shop.yourcompany.com and my.yourcompany.com. Now wouldn't it be a nice thing if, say, your company name was Nissan and you sold cars, you had a domain name like cars.nissan.com, or even better, nissan.cars.com. That way there's room for audi.cars.com, and nissan.computers.com, and even ford.tea.com.

This wouldn't have to take place overnight, but if a few significant industry domains could be purchased (then again, try prying news.com from c|net's cold, dead hands) the relevant subdomains could be rerouted to their more 'conventional' URLs.

If nothing else, it would be a nice service if, in addition to an index.htm (php, html, cgi, whatever) page and, often, a robots.txt page, there could be something like an others.html page that lists other companies or sites which might easily be confused with the domain name, along with links to those sites.

This would provide a 'near miss' functionality to web navigation, and code checking for the existance of an others.html document would likely see quick inclusion in several web browsers so that along with a 'search', 'security' and 'shop' buttons there would be an 'others' button.

Well, it's late and I'm rambling. I just wish big players (even people who I'd otherwise respect, like Sting) would stop trying to take things away from people by force. The net shouldn't work that way.

Comments? (3)

Though published in 1997, this report from the University of Calgary's GroupLab on how people revisit web pages is a fascinating piece on real people's activity patterns on the web. It's the most scientific look I've seen at an area usually dominated by anecdotal evidence. Take that, focus groups!

Comments? (4)

Or, more specifically, "What is the role of an Information Architect?"

A lot of people have been trying to answer this question lately. Whole books have been devoted to the topic. Still, I've found one of the oldest books in the genre, the Mythical Man Month, to give the best insights of what an information architect is and isn't. One reason I prefer this book over its more recent siblings is that it addresses the role of architect alongside the other roles, without a bias towards architect, strategist, implementer, or user. Another is that, having been written 25 years ago, it isn't trapped up in the current issues of the web. While some might see this as giving the book less relevance, I find that it provides less distraction, helping separate the core of information architecture from the idiosyncrasies of web development.

I'm still working on my definition of the role of information architect, culling from these sources and my own experiences, and I'll be posting it once it's complete.

Comments? (10)

Real world quiz of information architecture:

Extract meaning from the following:


0410001323
EXP:041002


While some context can be gleaned just from the bottom line, the information it provides is highly ambiguous (it could be interpreted at least 6 different ways). Adding the context of the upper line and the fact that it was found on a Crystal Geyser water bottle reduces it to about 2 interpretations, and that I bought it today in the United States from a vendor with high throughput pretty much narrows it down to one most likely interpretation (Expires April 10th, 2002), but it's an example of how people can overcome bad IA, not an example of good IA.

It's interesting to think how much effort went into designing the bottle and the label, and how little went into devising an understandable format for the expiration date.

Comments? (8)