adrian holovaty

August 30, 2002, 10:54 PM ET

Confusing breadcrumb navigation

Following up on my previous post, which lauded Computerworld.com's intuitive URL and breadcrumb navigation scheme...

Well. It seems many of the Computerworld.com pages have duplicated terms in their breadcrumb navigation. For example:

Home > Topics > Development > Development > Application Development

In this case, development is a subcategory of none other than...Development. Does this make any sense to anybody? If I'm reading that story and I want to get back to the Development page, which link do I click? Believe it or not, they do not reference the same page. It's obvious that having two levels of the same keyword is entirely nonsensical.

More examples: Security, Hardware, Software.

Comments (2) / Permalink

August 30, 2002, 10:37 PM ET

URLs and breadcrumb navigation working together

I was at Computerworld.com the other day when I noticed the site's elegant and usable URL/breadcrumb navigation schemes. In a Web full of gibberish URLs (such as this example), "smart" and hackable URLs are always a breath of fresh air.

On Computerworld.com, the URLs are written in plain English and convey a hierarchy -- for example, computerworld.com / departments / management / casestudies gets you to the "Case studies" page of the Management section, which is a subsection of Departments. Beautiful. When I loaded that page, I instantly knew where I was within the site tree.

And as if that weren't enough, Computerworld.com uses supplementary breadcrumb navigation that, on most content pages, echoes the URL word-for-word. Thus, in the above "Case studies" example, you'll see "Home > Departments > Management > Case Studies". This is an example to follow. (And it's very easy to do with Apache's mod_rewrite, which, in fact, I use on this very site.)

On a related note, I came across a good kuro5hin discussion on this topic, which linked to this smart-URLed version of the King James Bible. The Electric King James Bible, as it's called, lets you type a book, chapter and verse in the URL to get the respective Bible passage, like so: http://bible.conman.org/kj/Exodus.20:3-17. What a brilliant idea! There has to be an application for news sites here somewhere. Perhaps, to follow up on my recent post about easy-to-remember permanent links, a site might try something like this: newssite.com/us-china-trade/monday-wednesday might get you a page of all the stories about U.S.-China trade that were published between last Monday and Wednesday, inclusive.

Comments (2) / Permalink

August 30, 2002, 11:51 AM ET

Friday's links, and thoughts on table-based layouts

It's been a big week in browsers. We had a new Mozilla release Tuesday, and yesterday the final release of Netscape 7 was released.

American Journalism Review: Local TV management should get serious about the Web. They should also get serious about Web design; a lot of TV news sites leave much to be desired. (See this old site review.)

Jonathan Dube of CyberJournalist.net asked me to write a short piece on the making of the AJC Nursing Home Guide. That piece was posted online last night.

The influential Jeffrey Zeldman, author of To Hell With Bad Browsers, yesterday wrote that he finds table-based layouts acceptable, mostly because bad browsers haven't quite made it to hell yet:

When a popular current browser with otherwise good support for CSS chokes on something as basic as "float," and when millions still use 4.0 browsers (not always by choice), transitional XHTML layouts that include some table-driven formatting feel more and more like a reasonable choice.

My two cents: If content remains accessible in all browsers -- that's a must -- I still believe in advocating tableless, all-CSS layouts.

There are just too many benefits. I'm willing to live with having a site look plain in Netscape 4 because I believe Web users are more interested in obtaining information than being wowed by nice design. (The exceptions are sites that exist to wow the eye, such as photo blogs, promo sites for bands, or design/art sites.) An important distinction: I'm not saying users don't care whether things look pretty. Rather, they value the information itself more than the look of it. If a site is fundamentally easy to use, and uses proper HTML structure, it can get away with looking plain in old browsers.

Comments (1) / Permalink

August 29, 2002, 12:53 PM ET

Thursday's lunchtime links, and ideas on highlighting content

Luke Wroblewski: On the Pursuit of Simplicity. As I mentioned yesterday, news sites have much to learn about keeping things simple. (Link from InfoDesign)

New Architect: Weighing the Risks and Rewards of Standards.

Nathan Ashby-Kuhlman: The Web needs Xanadu features, particularly "the ability to address individual paragraphs, sentences or even words in someone else's document." I couldn't agree more. We need a way to permalink to anything on the Web (to borrow a blogging term).

On that note, I've recently been throwing around ideas with Stuart and Paul about devising some sort of standardized system for accessing a part of an HTML document. I have a rudimentary system already in place on Holovaty.com: You can add ?highlight=[highlightstring] to the end of any blog URL, and it'll highlight every occurrence of [highlightstring]. What I would rather have, though, is something like HighlightParagraph=[paragraphNumber], which would highlight an entire paragraph, or perhaps HighlightWords=[FirstWordNumber]to[LastWordNumber], where FirstWordNumber and LastWordNumber are the word boundaries of the document's requested selection (e.g., the 25th through 36th words).

The problem with such an idea, Stuart points out, is the user interface. It would be easy to program a publishing system to highlight the appropriate content, given those query strings, but we need something that would make it easy to create such query strings in the first place. In other words, in my last example, how would a user determine the word numbers of a selection without having to count them him or herself? An interesting problem.

Comments (7) / Permalink

August 28, 2002, 10:36 PM ET

No site review tonight

I've been extremely busy today working on an intensive high-school football database site for ajc.com (to be debuted Friday) and screening entries for the Online Journalism Awards. For that reason, I won't be able to post a site review tonight.

Comments (2) / Permalink

August 28, 2002, 12:50 PM ET

Wednesday's lunchtime links

In case you missed it yesterday, Mozilla 1.1 is out.

Digital-Web Magazine has started a new column, Keep It Simple, that aims to answer these questions: "Why do people make complex sites? When does a site become too complex? How should you judge whether a site needs to be complex? What techniques should you use to avoid complexity?" Was this column made for news sites, or what? Can't wait to see what the author, Peter-Paul Koch, has to say.

Steve Outing has a new Editor and Publisher column out today. It's all about Wi-Fi and Tablet PCs, and their effects on news publishing. Here's a snippet:

[The Institute for CyberInformation] is currently working on a digital-newspaper prototype for the Tablet PC...[The institute's director] says the prototype is designed to take a newspaper and make appropriate conversions -- initially using human editors and designers -- for presentation on a Tablet PC. It blends the best of newspaper and Web design into a new format.

I don't know the details of what they're doing, but it seems to me the aformentioned process would be unnecessary if news sites coded their content using Web standards in the first place. If a site is designed with accessibility and standards in mind, its content can be fed into any device capable of browsing the Web. By creating a "new format" for the Tablet PC, these folks are running directly against the Internet's movement toward a universal information-sharing scheme (XML). I'm not surprised Adobe is involved; I sense the company sees an opportunity to create a proprietary format -- much like Macromedia did when Flash was introduced -- over which, ultimately, it will retain control. (Again, I stress that I don't know the details. This is just a first impression.)

Also, Holovaty.com now pings weblogs.com each time I add a blog entry. In short, that means you can write a script that parses weblogs.com's changes.xml file on a regular basis and automatically determines the last-updated time of Holovaty.com and other blogs. (blo.gs offers additional features, in the same vein, that don't require programming.) Many thanks to Stuart for suggesting I set this up.

Comments (4) / Permalink

August 27, 2002, 1:23 PM ET

A blind computer user's thoughts on news Web sites

When I bring up the importance of Web accessibility with fellow online journalists, my concerns are generally brushed aside. "Oh, come on," people say. "We have much more important stuff to worry about. You can't really think there are that many disabled people out there."

Such an attitude is so hypocritical that I'm hesitant to call such people journalists. If they really cared about their "big picture" mission -- in most countries, providing the public with the information necessary to participate in a democracy -- journalists would bend over backward to make their content accessible to all readers. Instead, they don't seem to care. As I have argued time and time again in this blog, it's sadly ironic that the same news organizations that historically have trumpeted freedom of information tend to produce Web sites that make that information inaccessible. (Note: I'm not talking about making content accessible by publishing it for free online; rather, I mean the accessibility of the content itself -- the degree to which it can be understood by alternate browsers.)

But, let's be honest. It's understandable that online journalists wouldn't think about accessibility, given the relatively small number of Internet users who are affected by inaccessible sites. After all, who has actually met a user of the JAWS screen reader or a braille display? It's hard to keep those readers in mind if you've never heard from one.

Well, here's such a reader. Rich Caloggero of MIT Adaptive Technology for Information and Computing is a blind computer user. After he posted a few messages to the css-discuss list-serv, I asked him whether he would share his opinion on the accessibility of news sites for the benefit of designers who don't usually think about that. Mr. Caloggero was kind enough to answer my questions. I present the full interview here with the hope that Web designers -- whether journalists or not -- will gain a new perspective.

The interview

Q: To what do you attribute the inaccessibility of news sites?

Mostly, I attribute it to much too much information on one page, and the one-dimensional modality of speech. Just imagine that your favorite news site, say www.boston.com, had laid out with no formatting, that is, simply a long list of links, one per line, then the text of the article, if any, following all the links. How hard would it be to find something specific? What would browsing be like?

Q: As you noted in one of your postings, there's a lot of "garbage" at the top of each news story that makes reading news online frustrating. From a technical standpoint, what do you think is the best way for site designers to unclutter their pages? (For example, do you prefer a "skip to navigation" link, or would you just rather have the navigation at the bottom of the page?)

Firstly, I think all sites, not just news sites, should use structural markup wherever possible. For news sites, this is especially important, as it allows skimming by heading. If major sections of the page, say the beginning of the article in question, were marked up using heading markup, then one could use the screen reader to skip to the next heading. Perhaps lesser heading levels could be used for sections within the navigation and article body. The screen reader can be told to skip to next heading of a specific level, or simply skip to next heading.

However, all screen readers don't do this, and older versions most definitely do not. I believe that, if at all possible, sites containing large amounts of information, especially large numbers of navigation links on every page, should attempt to arrange things so that this "garbage" gets presented to a screen reader *after* the main content of the page. Said another way, the main content of the page should be the first thing one comes across on a page, especially a busy page such as one might find at an on-line news site or at any number of portal/community sites. Using style sheets, one can position navigation links wherever one wants, while having the HTML for the navigation appear *after* the content in the HTML data stream being sent to the client. From what I gathered from the CSS-Discuss postings I read, this is possible in modern browsers.

As a final resort, a "skip navigation" link should be included on pages which do not implement the previous methods. THe problem with this scheme is that it does not work when the page is dynamically generated, which is the case on many large news sites.

Q: Which news sites, if any, do a good job of being accessible? Which sites are particularly bad?

I really don't know much about news sites -- I just don't find them fun or interesting enough to bother. However, of the four that I've looked at -- www.boston.com, www.nytimes.com, www.yahoo.com, www.cnn.com -- boston.com was the worst. It seemed to have more untagged graphics than the others, and wasn't as well organized. Aside from this, no one stood out as being noticably more accessible than the others.

Q: Have you ever contacted a news Web site designer to point out how inaccessible a site is? How did he or she react? If you haven't, do you think it's worth doing at all?

I have not done this. If enough e-mail comes in on this topic, maybe they'd listen. However, I just don't see a few blind people's complaints making that much difference.

Q: Is the accessibility of news sites, and of the Web in general, getting better, getting worse or staying the same?

This is a difficult question. The optimist within says "sure, accessibility is much better than it was even a couple years ago." However, for every step forward we take, there are 3 new technologies just out which make sites slicker and more visually appealing, which are completely inaccessible to people using screen readers. I'm thinking right now about flash, shockwave, and other multi-media formats. When more and more people get high bandwidth connections, the landscape of the web will change drastically. Video, movies, interactive Java applets, and other things I've just not thought of yet will be confounding screen reader users far and wide.

Comments (11) / Permalink

August 26, 2002, 8:02 PM ET

News on the Semantic Web

Ftrain's Paul Ford, author of the excellent and much-talked-about How Google beat Amazon and Ebay to the Semantic Web, has a few ideas today on how news sites would fit into the Semantic Web. (The Semantic Web is a theoretical next generation of the Internet in which information and conceptual relationships between data are organized in a way computers can understand. But my definition doesn't do it justice; I encourage you to explore the topic for yourself.)

Ford's idea is for news sites to include uniform metadata (in essence, a consistent method of describing content) in each story. This metadata would be collected by a computer that would constantly be traversing the Web's news sites and storing its findings in a central computer. (Ford's theoretical example is "Newspurl.org".) Then readers could go to Newspurl.org, choose a topic, and get a list of all the world's news organizations' coverage of that topic -- ideally, sortable by date, publication, author, etc.

One of the benefits of this system would be that news stories would get permanent URLs. In one of Ford's examples, you could always go to http://newspurl.org/us-china-trade for the very latest news about trade between the U.S. and China. No need for fumbling around messy news site home pages that, as Ford points out, feature "near-nuclear war" one day and "a baby tiger born at the zoo" the next. We're talking incredible usability gains here.

I have several reactions to this piece:

First, news organizations will hesitate to do this because, in general, they don't like to share. Witness the recent deep linking hullabaloo. Witness the trend toward user registration. Witness the sporadic attempts at charging for content.

Second, why aren't news sites doing this on a site-by-site basis -- or even chain-by-chain basis -- right now? Take Ford's idea and apply it to a single news site. Make easy-to-remember, permanent links to repositories of stories on a single topic, and promote the heck out of them. Channel site traffic away from the home page and into these topic-specific index pages. If you're a chain, take it a step further and share content amongst your properties; God knows you've spent enough money on your content-management system, so use it to your advantage.

We're already seeing some of this. Washingtonpost.com does a decent job of sorting content into microtopics -- such as Espionage, Space exploration, even Biotech food. We're seeing it on blogs, too; publishing systems like Movable Type allow bloggers to assign categories to entries. (Here's an example of that.)

But these examples are the exception. Most sites plop news stories on their home page, grouping them with all other non-related events that happened to occur that day. Then, when day is done, the stories are lost into the dark depths of the Archives. Every day there's a new home page; every day users relearn the location of the stories they're interested in.

The biggest lesson of Paul Ford's piece is to organize content in an ultra-detailed manner and store it in a consistent place. The Semantic Web won't be here for a while, but news sites can lay the foundation for it by developing intelligent ways of tying stories together -- and convenient, user-friendly ways of presenting these associations.

Comments (2) / Permalink

August 26, 2002, 12:07 PM ET

Site news: Permalink URLs changed

I've changed the URL scheme for permalinks on this site. For example,

/blog/02/08/23/2328

...is now...

/blog/archive/2002/08/23/2328

I did this to make the URLs more readable and hackable. "2002" is more easily grasped than "02".

In the interest of avoiding a Knight Ridder-ish fiasco, I've played it safe: Every old link will still work.

Also, there's a new archive page that lists every entry in reverse-chronological order.

And, finally, the home page now only lists the 5 latest blog entries (not 10), in order to save you some load time. Older entries are accessible via the new archive page and the search engine.

Comments (5) / Permalink

August 26, 2002, 12:04 PM ET

Monday's lunchtime links

From last week: Zeldman says Show, don't sell when you're trying persuade bosses to use standards-compliant designs. Great reading.

Similarly, today the World Wide Web Consortium published a reader-friendly article, My Web site is standard! And yours? It includes a valuable set of rebuttals to common skeptical comments such as "I don't care about accessibility. People with disabilities are not in my target audience."

Mario Garcia examines newspapers' use of italics in this week's Eyes on the News. I once read somewhere that italics are a definite no-no in Web design because some Mac browsers render them illegibly.

Comments (2) / Permalink

August 23, 2002, 11:27 PM ET

Chicagotribune.com strike countdown

My buddy (and fellow Chicagoan) Justin gave me a heads-up about chicagotribune.com's baseball strike real-time countdown timer. It's on the site's home page, about halfway down, in a sidebar on the left side of the "Business | Technology" section. This is what it looks like:

This is a really cool, fun, useful and relevant use of JavaScript in a journalistic setting: It's a second-by-second countdown to the baseball players' union's Aug. 30 strike date. Will baseball team owners and the union come to an agreement by then? The clock is ticking -- for all chicagotribune.com readers to see. It gives the story a real sense of immediacy; plus, it grabs readers' attention more than a plain headline and blurb would.

However, as nifty as this feature is, I'd like to point out a mistake under the hood that ends up giving some users inaccurate information.

The problem is, the script depends on information on the user's computer to calculate the countdown time. Specifically, its JavaScript code checks your computer's system clock for the current time, from which it determines the remaining days, hours, minutes and seconds until Aug. 30. So, for example, if your computer's clock is a week slow, the chicagotribune.com countdown will be seven days longer.

Try it yourself. Here's how to change your clock: Windows / Mac / Mac OS X / Linux. For a good time, change your computer's clock to the year 2000.

See the problem? The Chicago Tribune's site is putting itself in a hazardous position by assuming users' clocks are correctly set. This results in journalism that's flat-out incorrect if users' clocks are incorrect. (From my experience, many are. But even if 99 percent of computers did have correct date and time settings, that remaining 1 percent would still get inaccurate information -- and that's unacceptable.) This is just as bad as misspelling a source's name or reporting a factual error. Don't think that just because this is the Web, journalistic principles can be compromised.

In short, I guess the lesson here is: Relying on client-side information to produce content that aims to be journalistically sound is a bad idea.

But this countdown timer idea is too good to be abandoned. It can be fixed, with the help of some sneaky coding. Instead of checking the user's computer time, the script should check the time of a resource it knows is always correct -- the chicagotribune.com server. Client-side JavaScript cannot access server information, so the solution is to insert server code within the JavaScript. Here's how to do that.

First, we need to determine the offset between the correct (server) time and the user's computer time:

// Get correct time from server
var correcttime = new Date("<!--SERVER TIME-->");

// Get user's computer time
var systemtime = new Date();

// Calculate offset between the two
var offset = correcttime.getTime()-systemtime.getTime();

Note that <--SERVER TIME--> should be replaced with whatever server-side date-generating code is available. Simple server-side includes will do the trick.

Then, use the offset each time the recursive showtime() function runs. Instead of today = new Date();, we'll use today = new Date(today.getTime() + offset); That's it.

I should mention that this will cause a delay of several seconds between the time the page is requested and the time the page is loaded fully (and the onload method begins). But, I believe a margin of error of a few seconds is allowable.

Comments (5) / Permalink

August 23, 2002, 12:53 PM ET

A glimpse into everyday life

Simon Hoegsberg's site, "Private the Public" (Flash required), is a collection of "snapshots of pedestrians who happened to walk into the frame of the camera" on a street in London. So normal, yet so fascinating. (Link seen on WebWord)

I'd bet a news site would be flooded with traffic if it started posting random, artful photos of normal people from around town. People would eat this up. Naturally, they'd would want to see whether they knew any of the subjects -- and whether they themselves were subjects. (Besides, people are on a big "reality" kick, what with reality TV and all that.)

Some might say this isn't journalism -- not to mention it might be a little Big Brother-ish -- but I think glimpses into the lives of normal people is a much more honest documentation of the day's events than the crap a lot of news organizations shove in front of our faces.

Comments (5) / Permalink

August 22, 2002, 11:55 AM ET

Coloring off-site links vs. in-site links

Mark Pilgrim explains how to make the Movable Type publishing system automatically display off-site links in a different style than in-site links. Is this something news sites should be doing? Would users benefit from a knowing which links take them to other sites? Rebecca Blood has done this for a while now, but I don't think it's particularly helpful on a blog, where most links are external anyway. And using three, maybe four, different link colors (in-site link, off-site link, visited link, link hover) on the same page could get confusing quickly. My opinion: This is technological overkill.

UPDATE, 2:45 PM ET: Steve Outing suggests this could be a good way of distinguishing between paid and free content. I still think things would get confusing quickly if sites started throwing around three different link colors; unless there were a Web-wide standard for these colors (which is unlikely), users would have to relearn the color scheme for each site that used it. Not to mention color-blind users and users of alternate browsers wouldn't be able to make the distinction. A good solution: Use images to distinguish paid content, as Economist.com does.

Comments (11) / Permalink

August 22, 2002, 11:48 AM ET

Thursday's lunchtime links

Online Journalism Review: Innovation in the Heartland -- An in-depth look at CJOnline, the site of the Topeka (Kan.) Capital-Journal, and all it's accomplished.

Interview with Derek Powazek -- The Web community guru stresses the importance of finding a niche. (Link from InfoDesign)

All about SmartPaper, potentially the newspaper of the future. (Link from E-Media Tidbits)

Comments (1) / Permalink

August 21, 2002, 11:38 PM ET

Site review: indian-express.com

An English-language news site based in India, indian-express.com is clean, colorful and easy to navigate. However, the site does a few things rather strangely...

• The first thing that leaps out at me is a drop-down box in the left navigation that calls for me to "Go to page [1]". Waiiiddaminute. Since when are Web pages numbered? I chose page 3, and suddenly I was in the "National Network" section. I assume these numbers correspond with a print edition of some sort? If not, I wish they were explained somewhere. (I couldn't find any help on this feature.) If so, this is a classic case of print mentality being applied to a Web site.
• On the other hand, this feature does deserve some credit for allowing users to traverse an edition of the newspaper in a logical order that (I presume) is determined by traditional editorial judgment. I think this idea is worth exploring; newspapers' linear nature makes for a much more focused reading experience. When I read a news site, a story will only "earn" my click if it appeals to me at face value (usually via a minimal tease, such as a headline). When I read a newspaper, I spend more time browsing stories -- perhaps because a physical effort is needed to turn the page, as opposed to a simple click?
• But, OK. Less theory, more nitpicky technicalities. As I've pointed out before, it's not a good idea to use JavaScript to load a page when a drop-down menu item changes. In this case, the "Go to page" drop-down menu would be better off with a "Go" button; users could make their choice and click the submit button to get the new page. Yes, it's one more click, but it makes the drop-down menu usable for IE keyboard users who would like access to more than just the first menu choice. (Not to mention the drop-down doesn't work with JavaScript turned off! At the very least, they could provide a NOSCRIPT message for users who have disabled scripting.)
• What's the difference between "Home" and "Front page"? (Both are listed in the left navigation under "INSIDE IE".)
• Every story I read had a horizontal advertisement that cut off near the bottom of my browser window. Some users might think the ad marks the end of the story. (I know I've made this mistake several times on other sites.)
• If there were such a thing as the Holovaty.com Headline of the Day Award, I would give it to this story.
• Here's a problem that many news sites seem to have: The Case of the Mysterious Date. Is the date in the upper-right of each news story (sample story) today's date, or the date the article was posted? I found out by changing the value of "content_id" in the URL's query string to a random lower number. I ended up on a story from March 2002, which listed the March date in the upper right corner. OK, now I know the date scheme -- but I doubt other, slightly less dorky, readers would go through that. A word or two, like "Posted:" or "Article written", would help. And maybe put the date inside the content area, instead of up there in margin land?
• On Holovaty.com, I try to avoid discussion of advertising and business models, but I can't help pointing out the "SERVICES" section of the site's left navigation. I find this an extremely good example of online advertising. When I visited the site, there were links to "FLIGHTS to India" and "MATRIMONIAL", both of which led to other sites. Seems to me this is a great way of giving users the advertising content they would actually be interested in seeeing -- without utilizing an annoying banner. (For you ethicists out there, I'll note the text is offset in a red color and is bolded. Other than that, there's no clear "advertising" label.)
• Intrigued by the about page's claim that the site was "[d]eveloped as per the international Internet access standards", I viewed the site in the text-only Lynx browser. Well, it looked downright lousy. Its front page (whoops, I mean "home page") was a mess of strange ALT attributes and nested tables. Mark Pilgrim would not be pleased.

Comments (2) / Permalink

August 21, 2002, 12:14 PM ET

Wednesday's lunchtime links

Digital Web magazine has started a new column about information architecture. The inaugural column gives a good history of information architects' role in the Internet as it exploded in the late '90s. David Wertheimer's column this month is also worth a read.

The folks at WebWord stumbled upon Holovaty.com and have been discussing my DHTML-based comment system from a usability standpoint. Some like it, some don't. What do you think? Does the comments box need a more obvious "close" button? Is the on-the-fly comment view annoying or hard to use? Right now, I've set it to dynamically load comments when there are four or fewer; otherwise, clicking "Comments" will load a new page. Has that confused anyone? I'm always interested in hearing what readers have to say.

Comments (10) / Permalink

August 20, 2002, 8:50 PM ET

How news sites keep robots away

After today's lunchtime links entry and the reader comments it brought about, I got to thinking about news sites' robots.txt files. Are other sites as robot-hostile as nytimes.com? I took a peek at a few news sites' files -- after all, if they're accessible to robots, they're accessible to humans -- and here are a few observations, along with links to the files themselves.

Robot-hostile sites

• Nytimes.com, as mentioned in today's previous blog entry, bans just about every file.
• Foxnews.com disallows every robot from everything! Coincidentally, David Gallagher wrote about this earlier today. (David points out that some Fox News content is indeed indexed in search engines, despite the disallowance.)
• Bostonherald.com bans pretty much everything.

Trends

• Many sites, such as post-gazette.com, hide their "contact us" pages. This is a good way of cutting down on (some) newsroom spam. (Not all robots follow the rules, however, which leads some to set intricate traps.)
• Other directories typically hidden are those that contain includes, advertising, images, JavaScript, story archives and server code (CGIs). No surprise there.
• Then there's the case of hidden directories for internal use, like "testing" and "development" and "temp". The number of sites that identified these directories surprised me; wouldn't sites be better off keeping them a secret by not linking to them, so the robots wouldn't have a means of finding them? Indeed, some robots -- the nasty ones -- specifically seek out directories that are banned by robots.txt. Pointing out those sensitive directories is like hiding from someone and yelling out, "Don't look over here!"

Random observations

• Boston.com's robots.txt file includes a note explaining robots "could cause some undesirable effects if they stumbled onto our voting scripts!", which leads me to recommend they modify their voting scripts to allow for only one vote per IP address per day.
• Not surprisingly, big chains tend to share robots.txt settings. Examples: Knight Ridder, Tribune Company, Cox.
• Dallasnews.com and the other Belo Interactive sites lack a robots.txt file. Projo.com does, however, have a very cool 404 page. (By the way, if you visit the Belo Interactive site with Mozilla, prepare to be insulted with a "Browser Not Supported" message.)
• You find some interesting stuff when snooping around robots.txt files. I found the URL for Villagevoice.com's server logs from Sept. 1996 to (interestingly) the day before Sept. 11, 2001. I also stumbled upon a page on azcentral.com that spit out some of the site's database table names. Somebody might want to fix that!

Comments (7) / Permalink

August 20, 2002, 12:51 PM ET

Tuesday's lunchtime links

Evolt.org: Describing Document Text for Accessibility -- "A key focus of accessible web site design is providing equivalent alternatives to auditory and visual content."

Joe Gregorio points out nytimes.com's robots.txt file (the file that delineates which parts of the Web site a robot is allowed to index) isn't very friendly. Namely, it bans robots from just about every file on the server. (More info about robots.txt.) Now that he mentions it, I've never come across an NYT story via a Google search. What a hostile policy!

Comments (6) / Permalink

August 19, 2002, 11:33 PM ET

Site review: thetriangle.org

The Triangle is the student newspaper at Drexel University in Philadelphia. It's an award-winning beauty of a site, obviously modeled after Salon.com but fresh nonetheless. I'd be hard-pressed to name a news Web site -- college or professional -- that makes better use of white space.

• Sitewide navigation is contained in two horizontal bars near the top. It's a great arrangement. Not only does this present the user with easy navigation decisions (as opposed to assaulting the user with dozens of links), it also frees up the left rails for inter-section navigation. For example, on an entertainment story page, the left rail has a list of all the entertainment headlines in the same issue -- with the current one bolded. This makes for easy perusal...
• ...as long as the headlines are descriptive enough. One common pitfall of such a scheme -- using headlines alone to index content -- is the feature headline, which doesn't really serve any purpose other than to be witty or "featurey." An example on thetriangle.org: Consumer Confidence. That headline may work in a print setting, or even in an online setting with the story right below it, but as a standalone navigation link, those two words tell me next to nothing. A good workaround for this problem is to have producers write a "long hed" and "short hed" -- one for the story, one for standalone links. Washingtonpost.com is one site I know of that does this.
• The site's style sheet is hard-coded into the top of each page. That's 6.47 KB of extra bandwidth wasted per page. The site loads quickly already, but the designers would do well to put that style sheet information in a linked style sheet, so you and I would only have to download it once.
• Speaking of the style sheet, the site's 11-pixel Trebuchet MS looks superb to these young eyes, but because the fonts are set using pixels, they're not resizable in IE for Windows, the world's leading browser. That means people like JD Lasica, with whom I just chatted today about this very topic, will just plain turn around and leave the site. Dive into Accessibility has a solution.
• The more sites I review, the more I realize how relatively unknown CSS shorthand is. In thetriangle.org's style sheet, font-family: Trebuchet MS, Tahoma; font-size: 12px; font-weight: normal; can be shortened into font: normal 12px 'Trebuchet MS', Tahoma;, and color: #000000; can be shortened into color: #000;. These might look like small changes, but they add up -- especially when you're hard-coding CSS into the top of each page, like thetriangle.org does.
• Bylines appear in a strange place -- the right rail, above advertising. Check out this page as an example.
• Right under the byline, there's an e-mail this story feature. But clicking it gives you a JavaScript popup window that says "coming soon". This is like selling a product that doesn't work yet; there's nothing to be gained for the user except frustration. It's like the dreaded "under construction" icon, but worse: Instead of being forthright about lacking the feature, the site presents the "e-mail this story" link as if it's fully functional. It's only after you click it that you find out it's not working just yet. There's a reason "under construction" pages have been banished to content hell.

Comments (1) / Permalink

August 19, 2002, 2:01 PM ET

Automated technology gone wrong

Forgot to mention something about the New York Times article I highlighted in today's lunchtime links. This article serves as an example of automated technology gone wrong. In the second-to-last graf, the word Northwestern (in "The Daily Northwestern") is linked to a company profile on Northwestern Corp. Clearly nytimes.com automatically links any occurrence of the word "Northwestern" to that company profile page. Silly inaccuracies like these damage a news site's credibility. An irrelevant link is just as bad as a misspelled word in a headline.

How would nytimes.com staff members fix such an error? Does their content-management system allow producers to bypass the auto-linking for a particular story? That'd be an easy fix. If not, I fear the only other possible solution, short of turning off the auto-linking sitewide, would be to rewrite the story so that the word "Northwestern" wasn't included. I doubt they'd go that far, but this is an interesting case of journalists being at the mercy of imperfect technology.

Comments (3) / Permalink

August 19, 2002, 12:07 PM ET

Monday's lunchtime links

Netscape 4.8 is out -- No, this isn't a joke. Yes, Netscape 6 and Netscape 7 have been available for a while now. Why someone would want to download a slightly different version of a bug-laden browser, I will never know. (Link seen on Zeldman and Webmaster World)

Jakob Nielsen: Let Users Control Font Size -- "Tiny text tyrannizes users by dramatically reducing task throughput." This is old news, but it's good to see everybody's favorite (or least favorite) usability expert chip in his two cents.

New York Times: Wired Students Prefer Campus News on Paper (name: cyberpunk21, password: cyberpunk21) -- "Even though college campuses are some of the most wired places on Earth, printed versions of college newspapers remain far more popular than their online editions." This is old news for recent college journalists like me. It's also a pity. Many college newspaper Web sites offer more features and are better designed than most professional news sites. I'll discuss one in tonight's site review.

Comments (1) / Permalink

August 16, 2002, 10:07 PM ET

Screenshot: NewsChannel2000.com icons

Since I'm usually busy Fridays, I'm going to start a weekly "screenshot" feature instead of Friday's site review. (Monday and Wednesday site reviews will continue.) Each week, I'll post a screenshot of a particular feature on one or more news sites. It'll be either a model I think should be followed or an example I think should be avoided. My comments might range from nitpicky to general; either way, I'll try to present features we all can discuss, learn from and be inspired by.

Today's screenshot is of the home page of NewsChannel2000.com, which redesigned this past week. It's their list of links that appears in the right-most column:

I want to point out two things about this list, one good and one bad.

The good: Those icons are styled with style sheets; they're not hard-coded in. This is accomplished here by wrapping the link in a DIV tag and setting the DIV's background image to the icon. Here's the relevant code for the first item in the list:

<div class="computerlink"><a href="http://www.link.com/" class="iconed">Get Cool WESH Merchandise</a></div>

The div class="computerlink" corresponds to this piece of CSS code:

.computerlink {background-image: url('http://images.ibsys.com/2001/0126/435589.gif'); background-repeat: no-repeat;}

Note that this can be rewritten to background: url('http://images.ibsys.com/2001/0126/435589.gif') no-repeat;.

The background-image attribute sets the icon, and background-repeat is set to "no-repeat" so that the icon only appears once. Otherwise it would be tiled under the link repeatedly, looking like this:

The class="iconed" code pushes the link text over to make room for the icon. Here's the relevant CSS:

.iconed { padding-left: 22px; }

Without that left margin, the text would overlap the icon, like so:

The advantage of using style sheets this way is: If the NewsChannel2000.com producers ever want to change that computer image, all they'd have to do is change one CSS file to make the change sitewide. Plus, it saves bandwidth, because the code is more compact.

The bad: OK, so the icons are styled in a relatively innovative way. But what the heck do they mean? What's the difference between the computer, the file folder and the page with a corner bent? And what's that globe pointing to? The question mark is obvious, as is the envelope, but the rest of the icons are blatantly unusable. This is an entry for the Interface Hall of Shame if I ever saw one. In the name of fairness, I tried to find a site FAQ page of some sort, but I couldn't find out, anywhere, what those icons mean. If anybody has any ideas, please post a comment and enlighten us all.

Comments (3) / Permalink

August 16, 2002, 12:47 PM ET

Friday's lunchtime links

A List Apart: 10 Tips on Writing the Living Web -- Everyone's talking about this article. Not necessarily design-related, but it definitely applies to online journalism.

Interview with Curt Cloninger -- The author of "Fresh Styles for Web Designers" says "beauty enhances usability" but "HTMinimalism is a style you can use for a newspaper site." (Link seen on WebWord)

New Architect: Making Mistakes Well -- "By improving contingency design, online businesses can help customers recover from mishaps, while increasing conversion rates in the process." In plain English: Improved error pages can and will lead to happier customers. (Link seen on WebWord)

Comments (1) / Permalink

August 15, 2002, 12:54 PM ET

Thursday's lunchtime links

SitePoint redesign -- Just for kicks, Simon Willison has redesigned SitePoint, a popular design site, using an all-CSS layout. This is proof positive that even complex layouts can be duplicated using style sheets. (Also see Steve Clay's Amazon.com redo.) News sites should step up to the plate. (On a related note, good discussion continues in the comments to my Why news sites don't use XHTML and CSS post.)

Washingtonpost.com now officially requires users to provide personal information, which is stored in a cookie, in order to access some stories. From what I could tell, it started yesterday afternoon/evening (Eastern Time). But which stories bring up the registration prompt? As far as I can tell, only metro stories. Other stories (politics, health, etc.) remain "free" for the time being. (Also see my previous blog entry.)

Nathan Ashby-Kuhlman: Technical restrictions hinder customization -- "Why is it that people like me can run server scripts and databases on our personal sites for less than $100 a year in hosting fees, but producers at sites with huge budgets and dedicated datacenters are stuck with static HTML and no control over content management systems?" I've wondered the same thing, brother. I've wondered the same thing.

Comments (6) / Permalink

August 14, 2002, 11:28 PM ET

Site review: ljworld.com

As I mentioned earlier today, the Lawrence (Kan.) Journal-World's Web site has made a few design changes. Now, it's cleaner and prettier. This site's one to watch, because it recently acquired a few high-ranking staff members from the award-winning CJOnline.

Before I go further, here are examples of the old design, for comparison:

• Dec. 6, 2001, story (HTML)
• May 19, 2002, story (PDF; Acrobat required)
• June 22, 2002, story (HTML)
• Old CSS file

As always, here are a few comments and suggestions on the changes:

• Probably the most dramatic change is the addition of a left rail. If memory serves me well (and if the old design examples are accurate), there wasn't one before; navigation was relegated to a somewhat unattractive image map. Clearly the left rail is a tremendous improvement.
• I find it interesting that the left rail's section choices are listed in alphabetical order. This is something you don't see often on news sites. As a result, it's a bit easier to find a section you're specifically looking for, but it's more difficult to browse freely, because there's no built-in editorial judgment that places one section above the other. (For example, nytimes.com places "International" first, followed by "National", "Politics" and sections of either lesser importance or lesser popularity.) It can be argued that the alphabetical method helps users find what they want quickly, but at the same time, it can be frustrating. Indeed, a Wichita State University usability study of sitemaps concluded: "it was more difficult to find information in the Alphabetized sitemap because they had to guess how this information was worded in the menu." An ljworld.com example: Try finding "World". Give up? It's next to "Nation".
• I keep thinking that dark blue horizontal bar under the site's logo is breadcrumb navigation. It's the slashes that do it; vertical bars would erase confusion.
• So simple, but so effective: I love how many pages have a large, red section title in the content area. (See this page for an example.) These red blurbs existed on the previous site design, but now they're more pronounced. Bigger, but not too big.
• This feature was on the old design, too, but it's still worth mentioning: Stories' first six words are bolded. That's a nice visual pointer to the start of the story, but I'm afraid it's another case of automation technology and journalism butting heads. Like HarkTheHerald.com's photo automation (discussed on my site a few weeks ago), automatically bolding the first six words of each story is like driving with your eyes closed; you never know what might happen. In a quick glance over the site with this in mind, I found one example right away: the Births section, where the first two parents are bolded and the rest aren't. On this Births page, what makes Erin Myers and Brian Franklin more special than the other proud parents in the paper today? Readers notice these things.
• Very few links are clearly designated as such. Sitewide, only a handful of links are underlined (not counting mouseovers), and many links are "set off" in a color only a shade or two different from their surrounding text. I found myself hovering my cursor violently around pages, waiting with bated breath for the arrow to turn into a hand. In some cases, such as headlines on the home page, it's excusable; in others, such as reporters' bylines and the columnist archives, the color difference is so slight you end up feeling like you won a game each time you stumble upon a link. And if I, blessed with decent eyesight, have problems, imagine how color-blind users must feel. (Vischeck is a good tool to use in testing what your pages look like to color-blind users, and dive into accessibility has much to say on the topic.)

Comments (1) / Permalink

August 14, 2002, 12:13 PM ET

Wednesday's lunchtime links

Steve Outing's "Stop the Presses": Examining Paid Content's Future -- Points out five trends in the online news industry. They're all depressing and/or disturbing except number 4.

Nathan Ashby-Kuhlman: Useless URLs at two papers -- Nathan points out blatant usability failures at statesman.com and ocregister.com.

I Want Media: Interview with Chris Schroeder -- Great ideas, foresight and advice from the CEO and publisher of Washingtonpost.Newsweek Interactive.

And speaking of washingtonpost.com, today the site is set to make the leap to requiring user information. (See my previous blog entry on this topic.) As of this writing, they still haven't flipped the switch; stories are still accessible without registration.

Interview with Gerry McGovern -- McGovern, content-management expert, says "we need designers who think like editors." Some hackneyed ideas here, but it's worth a read. (Hat tip: Small Initiatives)

Rob Curley's only been there for a few weeks, but the Lawrence Journal-World already has changed its design. It's a substantial improvement that bears some resemblance to Rob's former site, CJOnline -- no doubt because the same lead designer worked on both. (CLARIFICATION, 7:02 p.m.: I'm told the same lead designer didn't work on the ljworld.com redesign.) More on this in tonight's site review.

Comments (1) / Permalink

August 13, 2002, 12:45 PM ET

Redesign at NewsChannel2000.com

NewsChannel2000.com is redesigned, according to this press release. It's the site for Orlando's NBC affiliate. They've abandoned a left rail in favor of horizontal navigation. More comments on this later.

Comments (1) / Permalink

August 13, 2002, 12:44 PM ET

Tuesday's lunchtime links

CNET: Language barriers on the Web? -- XHTML 2.0 will not be "backward compatible", which will likely leave millions of Web pages incompatible. One editor of the XHTML draft says: "There's going to be a cut-off point...If we're going to move the Web to XML, we've got to move it."

Salon: The media titans still don't get it -- Much of this article is blah, but the last two sentences are worth memorizing.

Comments (1) / Permalink

August 13, 2002, 12:40 AM ET

Smarter search results

No site review tonight, because I've been busy with a few new Holovaty.com features.

Both features have to do with helping people find what they want. I've programmed my search engine so that result pages highlight your search terms -- each word in a different color, just like Google. For example, try a search for auto refresh home page and check out the highlighted terms.

Of course, sometimes you just plain don't need text highlighting. That's what the aptly named "Remove highlighting" link is for.

If that isn't exciting enough, here's the coolest part: Following webmasterworld's lead, I've added a Google-specific highlighting feature. If somebody lands deep within Holovaty.com via a Google search, that person's Google search terms will automatically be highlighted on my site. (And, again, there's the option to turn them off.) Plus, there'll be a link to search only Holovaty.com for the search terms -- an improvement on webmasterworld's system.

Eventually I'll add support for Inktomi, AlltheWeb, Altavista and maybe a few other search engines, but Google is the most important, in my opinion.

If news sites did something like this, people would consider them a heckuva lot more useful. I recall hearing an absurdly large percentage of hits to news sites come via search engines. Why not lend a helping hand to potential readers?

Comments (6) / Permalink

August 12, 2002, 7:00 PM ET

Attempting an all-CSS layout at hollandsentinel.com

In a post on the css-discuss list-serv, Aaron Schaap of The Holland (Mich.) Online Sentinel said he'd been having a tough time redesigning his site completely in style sheets. After getting a tip or two from the list, he decided to remain using a table-based layout. (His boss had required the new design work in Netscape 4.7 -- the bane of all Web designers' existence.)

Well that's demoralizing. I agree with Darrel, who had this to say:

How much of your logs are NN4? If it is a small percentage, I'd say just kill the CSS layout completely in NN4 and let the content display as-is. I've found going completely to CSS formatting is actually a benefit for NN4 as you can kill the CSS layout for NN4 and the page is still rendered in a very accessible, readable way in NN4 (with careful planning, of course).

Since you are a newspaper, most people just want to read the articles anyways, and NN4 will let you do that.

Comments (3) / Permalink

August 12, 2002, 12:33 PM ET

Monday's lunchtime links

CSS-discuss list archive -- One of my favorite mailing lists has now made available publicly a Web-based, searchable archive of every list posting ever. This is a Mecca for CSS tips. Got a CSS question or problem? Chances are somebody else has brought it up here.

Why Usability Matters -- Poynter's Monica Moses explains how we can "make news content look easier, more scannable, less a part of the numbing cacophony of the daily media barrage." Intended for print designers but applicable to the Web, too.

Everything I need to know about building Web pages I learned on pre-Kindergarten sites -- Small Initiatives has this friendly list of Web-building advice from a child's perspective. Remember: "There's something wrong if a user needs help from Mom or Dad."

Comments (1) / Permalink

August 10, 2002, 1:44 PM ET

A new online news approach

Boing Boing offers a cool idea -- a "tabbed newspaper" using Mozilla's bookmark tabs. There's some good discussion, too.

(By the way, Mozilla rules.)

Comments (2) / Permalink

August 10, 2002, 1:41 PM ET

Are 99 Percent of all Web sites obsolete?

The answer is yes, Web guru Jeffrey Zeldman says in a new interview. Read this if you still don't think following Web standards is essential. Read it anyway if you're already enlightened. A snippet:

Most of us are still using techniques that date from the mid 90s. We slice and dice images, stick them in tables, write scripts for different browsers, use browser detection to find out which user agents are being served, and so on. We're writing markup with no basis in structure, because when we started writing h1 and h2, Netscape put big white space under our headlines, ruining our pretty layouts. So we learned to write nonstructural stuff, such as div class="subhead2". We couldn't use HTML out of the box, so we learned stuff that made our sites look OK. But we robbed them of any underlying structure.

Comments (1) / Permalink

August 9, 2002, 1:34 PM ET

Lunchtime links

BBC: Why ads on the net don't work -- "The solution may be for the web merchants to stop acting so desperate and instead to respect site visitors and try to give them stuff they find useful, rather than irritating." Hallelujah, brother. News sites need to listen up.

RichInStyle.com -- An outstanding resource I stumbled upon today in my efforts to convert ajc.com's home page to an all-CSS-based layout. Especially useful are the browser bug reports.

The Second-Level Digital Divide of the Web and Its Impact on Journalism -- Overly academic but somewhat insightful.

Mario Garcia on centering headlines -- Any news sites do this? I can't think of any off the top of my head.

I won't be posting a site review this evening; I'll be at a Tom Petty concert.

Comments (3) / Permalink

August 8, 2002, 1:23 PM ET

How empirical is Web design?

Jay Small's latest Sensible Internet Design newsletter is out. A great read. It mentions Jay's "Range of Design Specialties" chart (slide three in this presentation), which places Web designers, abstract artists, programmers, infographics artists and print designers on a spectrum ranging from "emotional" to "empirical".

My take: Web designers should be placed a little further toward "empirical" than where they currently stand (near the middle). I say this because a goal of Web design is not only to make things look pretty, but also to make them accessible to all types of browsers -- unquestionably an empirical task. This goal is seldom achieved by most sites -- particularly the sites of news organizations, which, ironically, claim to fight for freedom (and, it follows, accessibility) of information.

This aspect of accessibility is a focus almost completely lacking from print design, where medium always stays the same. In contrast, the Web is a medium of "sub-mediums": One person's Internet Explorer browser displays content much differently than another person's text-only Lynx browser, which in turn acts differently than the browser in a wireless Internet device.

But designers from the print side often neglect to consider this essential point. Instead, they bastardize the Web, narrowing its reach and cheapening its value by applying their print mentality to a very different medium. Thus we see Roger Black, a leading print designer mentioned in Jay's newsletter, creating sites, such as his very own, that look lovely in the latest browsers but contain absolutely no non-navigational content in text-only browsers. (On Black's own site, in a ridiculous move, all content is in ALT-tag-less images.)

Yes, making things pretty is important. And yes, many print design concepts are valid on the Web, to some degree. But I stress Web design is an empirical art.

Comments (4) / Permalink

August 7, 2002, 11:24 PM ET

Site review: Nzherald.co.nz

Farah Iqbal, the designer at The New Zealand Herald Online, told me that site changes are in the works and suggestions would be appreciated.

Let's all help out. I'll go first; others, feel free to post comments below.

• The first thing that comes to my attention is the large banner ad on many of the pages. It measures 760 by 120 pixels, and it's a whole lotta trouble. Not only is it distracting, it actually slowed my (relatively fast) computer down -- the MP3 playing at the time skipped, and there was a delay in scrolling down the page. This isn't a weblog for online business, so I don't care to debate the financial merits of such an ad, but I strongly urge that this monster be tamed. Having an ad slow down one's computer is annoying and might very well send users elsewhere. One suggestion: Make it a one-frame ad, not animated.
• Another disadvantage of the ad: Because it's so gigantic, I perceive it as an almost-integral design element of the page. For this reason, the pages that lack the giant ad threw me for a loop. Moving from giant-ad page to no-ad page can be very jarring. As I surfed the site's first few pages, my eyes learned to ignore the ad and begin scanning pages from about a third of the way down. But when I came upon a page that didn't have the ad, it was an uncomfortable change -- like having the rug pulled out from underneath me. Consistency in placement of page elements, including advertising when it's this big, is important.
• The far-right rail has a similar effect. Sometimes, it's used for vertical ads; at other times, it's used for sidebar information. I have a feeling many users ignore the sidebars because they're in (what users perceive to be) the ad rail. Advertising staffs can say all they want about forcing people to look at ads by placing them in editorial space, but I believe users make quick judgments about what is editorial vs. advertising space and learn to ignore the latter. I think it'd be a better idea to put related content within the content area, but I'd love to hear more opinions on this.
• Usability of the top-left drop-down menu can be increased by eliminating its JavaScript onchange event. Right now, when I choose a menu option, it immediately takes me to that page immediately, without my having to click "GO". It's great that the "GO" submit button is provided for non-JavaScript browsers, but in some cases (mainly, when using the keyboard to browse), it's impossible to make a choice other than the top link without being redirected. (See Accessible Scripting's informative explanation.)
• All fonts are set using absolute widths, making it impossible for Windows Internet Explorer users to resize them. (Although a workaround is given on the site's outstanding FAQ page.) Dive Into Accessibility has a great page on this topic, although its solution is a bit too much of a hack. My take on the matter: Use ems instead of pixels.
• As mentioned before on this site, it would be a good idea to use correct structural markup, such as H1 and H2 tags, for headlines, bylines, etc. Such markup gives text-only browsers a better idea of what kind of content they're displaying. (And that's just one of the advantages.) Also, I encourage heavier use of style sheets in order to separate design from content. Quick tip: The stylesheet's current font-family: arial, helvetica, sans-serif; font-size: 12px; line-height: 19px; font-weight: bold; can be combined into font: bold 12px/19px arial, helvetica, sans-serif, and color: #000066 can be shortened to color: #006. Might not look like a lot, but it saves a lot of bandwidth and cuts page load in the long run. (Here's a useful page on such CSS shortcuts.)
• It's always good to see breadcrumb navigation! (This is a topic previously mentioned here.) And section names in the left rail are highlighted according to the page you're on. As a result, the site is easily navigable, and that's commendable.
• The search form is featured very prominently, in the middle of the page and above the news content. It's obvious that making site search easy should be a high priority, but here, the form's dead-center placement almost regards search too highly. As a recent Digital-Web Magazine article explains, searching should be a last resort, not a default method for retrieving content. The key is making navigation prominent and usable. The search form should still be easy to find, just not this easy. For what it's worth, Jakob Nielsen has recommended search forms be in the upper right corner of a site -- sufficiently out of the way, but instantly apparent when users need them.
• Speaking of site search, what do the percentages next to the search results mean? (See a sample search.) The site's FAQ offers this: "[R]esults are ordered by the relevance of the story to your keyword." Yes, but how are the percentages calculated? They don't mean anything to me unless I know.

I hope that's helpful. Everybody, chime in with your own comments.

Comments (7) / Permalink

August 7, 2002, 12:21 PM ET

Lunchtime links

404 Research Lab -- History, examples and more.

typoGRAPHIC -- All about typography. "Hip" capitalization, too.

Webmonkey: Web Standards for Hard Times -- Why you shouldn't ignore Web standards.

E&P: NYT adds Google search -- A smart move.

Digital-Web: Building a semantic Web site -- A readable look at the future of CMSes.

Comments (1) / Permalink

August 7, 2002, 9:11 AM ET

Washingtonpost.com to ask for personal information

The Wall Street Journal reports washingtonpost.com will begin asking users for gender, year of birth and zip code, starting today. It'll be voluntary until Aug. 14, when users will be required to enter the information to access news stories, the article says. Information will be stored in a cookie.

It'll be interesting to see how they pull this off, especially from an accessibility standpoint. What about users who have disabled cookies in the interest of privacy? Clearly they'll have to enter the personal information every time. What about Lynx users? Lynx has the ability to accept and save cookies, but they're not stored between browser sessions.

UPDATE, 7:47 PM EST: Here are some more resources:

• Reuters article
• Washingtonpost.com press release

Comments (1) / Permalink

August 6, 2002, 9:39 PM ET

Syndicate this site

A few days ago I set up an RSS feed that syndicates the 10 latest Holovaty.com blog entries. Today I've added custom feed capability that syndicates XML of only blog entries that contain a word or phrase of your choosing. More information is on my RSS feed page.

Comments (1) / Permalink

August 6, 2002, 12:24 PM ET

News site correction pages

Ace ajc.com intern Nathan Ashby-Kuhlman devotes a blog entry to how different news sites publish corrections. As a complement, here are a few additional corrections pages, good and bad:

• Honolulu Advertiser -- Has a nice corrections page with links to the corrected stories. Corrected stories themselves are updated and flagged as such at the bottom of the page. Outstanding.
• CNET -- An intensive corrections section, sorted by story type. Corrections are made to the original story.
• Seattle Post-Intelligencer -- A list of corrected stories since early May. The corrections themselves aren't explained on this page, but the correction is made in the body of the story. A great system, I think; although it'd be a bit better if they listed the actual corrections on the index page, too.
• Minneapolis Star-Tribune -- Lists all corrections since October 1999. But corrections themselves don't link to the stories; instead, they identify which print-edition page the story appeared on. (Funny. I thought this was the Web.)
• Baltimore Sun -- Nice to see 'em in one place, but links to the articles would be nice.
• El Paso Times -- "The last name of the youngest El Paso Community College graduate was incorrect in the El Paso Times May 10." This is frustrating. Which story?
• Sacramento Bee -- In case you didn't get them the first time, they're attempting to "set it straight."

Comments (1) / Permalink

August 5, 2002, 9:15 PM ET

Site review: HarkTheHerald.com

HarkTheHerald.com, the site of The Daily Herald in Provo, Utah, makes up for its substandard design with a handful of innovative, mostly database-driven features. Some thoughts:

• The site looks and reads like a blog, and that's a good thing. The home page and section fronts simply list all stories in their section (from what I can tell, the home page serves as a replica of the print edition's front page), with a headline, byline, date, short blurb and a few links: "Read More", "comments?", an icon for a printer-friendly version and an icon for "send this story to a friend". Can they get more blog-like? I thought it was a blog the first time I visited the home page. It might not be sexy, but its usability is outstanding.
• The horizontal navigation bar is a list of the days of the week, starting with Sunday. The current day is highlighted, and the other days are linked to their respective content, which means a user always has seven days' worth of content immediately accessible. It's simple to use, but, because the days are always listed from Sunday to Saturday, the days get counter-intuitive. For example, when I visited the site today, Monday was bolded; but clicking the Tuesday link, which is directly to the right of the Monday link, took me to last Tuesday's content. It's obvious that a news site can't print tomorrow's news, but the way this navigation is presented suggests that.
• As far as I can tell, most of the content on this site is from the print edition. Therefore I assume the site doesn't change much throughout the day. Why, then, is auto-refresh set on the home page and every section front? (See my previous post on this topic.)
• Users are encouraged to post comments on each story; comments are embedded at the end of each article. This is a great move on the site's part, a feature that other news Web sites wouldn't touch with a ten-foot pole. My only beef is that a user has to click "Post a comment" to get a new page that prints the article all over again (strangely, in a different font size), without the previous comments and with a form for posting. Why not put the form right on the first page, right under the other comments? Clicking to bring up another page is an unnecessary step.
• All the photos on HarkTheHerald.com seem to be in a database, which leads to some pretty cool -- but possibly controversial -- stuff. Photos in articles are only thumbnail versions; clicking on a photo brings you to a larger version with caption. From that page, you can click "previous image" or "next image" to get another image from that particular issue. (What determines "previous" or "next"? Probably just the order in which they were entered into the database.) It's a cool, sticky way to get people clicking. Plus, "back to thumbnail view" takes you to a thumbnail page of all the photos in that particular issue -- again, in no significant order. Thus, at a glance, you can see all the photos from a single day. (Everything's powered by an open-source PHP script that anyone can download.)
  
  It's a very cool feature, but it presents some journalistic pitfalls. Namely, by lumping all an issue's photos and comics together on a single page, the chance of inappropriate combinations is high. I pondered this possibility this evening, and, sure enough, on today's photo index page, an obituary photo of a dead man sits directly next to a photo of a Navy SEAL aiming a gun. Utterly tasteless:

• In addition, from the photo detail page, it would be nice to see which story a photo goes with. Cross-linking stories and photos can't be too difficult.
• Users can also easily navigate the photo archives using a breadcrumb navigation scheme at the top of the photo detail pages. One example of computer logic winning over human logic, though: On the year index pages, month names are listed alphabetically instead of chronologically. It made me do a double-take, several times.
• The site's staff page is a model to be followed. Each staff member gets an extended bio, photo and list of favorite links. That's followed by an e-mail address and a link to a Web-based contact form for people who don't have e-mail access. Then, to top it off, a phone number. Some sites don't want you to contact their reporters. Others, like this one, are much friendlier.
• Finally, there's a slick handheld version -- and its story pages double as the printer-friendly versions for the normal Web stories. Great reuse of content.

Comments (7) / Permalink

August 2, 2002, 11:10 PM ET

Site review: gazette.com

Gazette.com, the site of the Colorado Springs Gazette, is a remarkably easy, pleasant read. It's clean, it's easy on the eyes, it's fast-loading, and it relies heavily on Verdana. All good attributes in my book.

• The site makes a sincere effort at using correct HTML structural markup. Headlines are H1s, unordered lists are ULs, and much of the text styling is done in CSS. This is laudable; users of text-only browsers and other alternate browsers will be thankful. Unfortunately, though, the CSS is hard-coded into the top of every page, which means users have to download the same code all over again for each page they visit. The site's producers would do well to link an external style sheet to each page; then users would have to download the styles only once and let their browser's cache handle the rest. Still, it's good to see structural markup.
• Oooh, and the left-rail navigation links all have TITLE attributes, too. And the tags provide useful information! This is a rarity in the online news world.
• Obits get their own navigation link in the left rail, but the link takes you to the middle of the metro page, from which you have to follow another "Obituaries" link to get to the obits page proper. That's one too many steps. I assume they're doing it this way because the URL of the obits page changes daily, and they don't want to change their sitewide navigation each time around. A solution might be to set up a redirect and edit that each time there's a new obits page.
• The p.m. edition, which is a Web-only bunch of stories published in the afternoons, is clearly labeled on the home page with a huge icon. (Well, maybe it's a little too huge, at 13K; and besides, it should be a GIF.) I can't stress enough how convenient it is having all the p.m. edition stories in one place, clearly labeled, at the top of the home page. And gazette.com takes it one step further: All the p.m. edition stories are published on a single page. Normally I'd question having unrelated newspaper articles together in a single Web page, but here it seems to make sense. The articles are short enough, there aren't too many of them, and I have a feeling they're more likely to be read because they're all in one place. A bright idea.
• The site seems to link heavily to MSNBC. OK, so they have a "strategic agreement." But the links aren't so strategic. For instance, when I clicked on the "msnbc" link from the left rail on the home page, I was taken to the normal MSNBC home page, where a popup box prompted me for my zip code. Wouldn't it be a good idea to assume people are coming from the Colorado Springs area? Instead of asking for the user's zip code, the prompt could list a few zip code suggestions from Colorado Springs (with, of course, a backup input field in case the user isn't from that area) and make the transition much smoother. I'm sure the "strategic agreement" generates some money based on some wacky business model, but I question whether having the link to MSNBC is, in any way, helpful at all to the user, who's the person who really matters most.

Comments (1) / Permalink

August 2, 2002, 11:35 AM ET

MSNBC in early stages of redesign

A piece about MSNBC in Online Journalism Review reveals the site is in the beginning stages of a redesign, which possibly will be unveiled in the first quarter of 2003:

[General Manager Scott] Moore promises the site "absolutely will remain distinctive." He offers Slate as an example. "MSNBC is going to maintain its own unique brand and the design is part of that brand ... On the other hand MSNBC's design has not changed in 5 years. It's time for a refreshing of the design and the user interface. We're not going to strip it down."

Comments (5) / Permalink

August 1, 2002, 7:36 PM ET

Why news sites don't use XHTML and CSS

A reader who noticed the "XHTML and CSS" note at the bottom of my site e-mailed me asking why most top online news sites haven't made the switch to XHTML code and CSS-driven layouts. I think they haven't done so for the following three reasons, none of which is acceptable:

• On average, many Web producers at news sites are slow to adopt new technologies and/or don't want to change their work habits. Although learning XHTML and CSS is easy -- I daresay even easier than old-school HTML -- producers don't want to learn it because that would mean learning something new. Online news producers are journalists, not programmers or Web designers. They care about breaking news, not necessarily about Web standards or clean code. I don't mean that as an insult; I'm just telling the facts. There's an unwillingness in the news business to adapt new technologies right away -- even when the technologies would improve products tremendously. (Of course, there are exceptions to this; I'm just echoing what I've seen and heard from others in the field.)
• Many Web developers are simply unaware of Web standards, XHTML and CSS-driven layouts. The Web Standards Project's FAQ page is a good place to start, as is the New York Public Library's Online Style Guide.
• Webmasters are scared of losing readers who use old browsers that don't support this technology. This point is somewhat more l