NATHAN ASHBY-KUHLMAN > Blog entries from October 2002

Wire feeds in the wrong time zone

I’ve already picked on Knight Ridder recently, but I just discovered another problem with their centralized content management.

Many of the Knight Ridder sites carry a certain AP feed that I assume is centralized. The trouble is that the timestamps are also centralized. Papers in the Eastern time zone display timestamps in “PDT (or probably PST by the time you read this). Some examples:

Other sites use a Reuters feed, in which the timestamps are in Eastern time, regardless of where the newspaper is based:

Programming a separate version of these wire service pages for each time zone Knight Ridder serves would be a piece of cake. They should be ashamed of themselves for using time zones outside of each newspaper’s geographic area, confusing readers and diminishing their local brand identity.

Update October 26, 2002, 11:58 pm: Those sites I listed as using the PDT AP feed are now apparently using the EDT Reuters feed on their front pages, although the links I gave above to inside pages still show the AP feed. Perhaps Knight Ridder changed all of its sites over from AP to Reuters between early afternoon, when I first noted this issue and copied down some URLs, and this evening, when I wrote the blog entry?

POST A COMMENT on “Wire feeds in the wrong time zone”

Your name: (optional)

Your site: (optional)

Comment: (only <a>, <em> and <strong> allowed)

Breaking news, delayed by the CMS

The content management system at work requires a few minutes from the time a producer finishes working on a page until the changes actually show up live on the site. The CMS at the paper I interned at in summer 2001 required 20 or more minutes.

The designers of systems like these obviously do not understand that breaking news should appear immediately, as soon as a producer finishes with it, not at the system’s leisure. Radio and TV don’t buffer their news programs for five minutes before the viewer sees them — live news means live, delivered at the speed of light. Print newspapers do buffer their news for hours between the last deadline and when papers are delivered. If newspaper Web sites are supposed to complement their print editions by being continuously and instantaneously updated, they should not be creating publishing systems with a used car salesman’s definition of instantaneous.

The technical reason for these delays, I’m assuming, is that the CMSes cache pages they produce. This is a smart idea for reducing strain on servers, because if hundreds or thousands of people request the same story, the CMS shouldn’t have to rebuild it from scratch each time. Rather, these CMSes check over their databases every few minutes, looking for articles producers have changed and updating the cached versions that visitors actually see.

My quarrel is not with the idea of caching but with its implementation. When a producer edits a story, the CMS programming that saves changes into the database should update that story’s cache immediately. The CMS should not wait a few minutes until its next regularly scheduled check for updated stories. Breaking news should not be hidden from readers for a few minutes just because a Web circa 2002 content management system can’t duplicate the Web circa 1994 ability of FTP to update immediately.

Comment by shayne, posted October 26, 2002, 12:43 pm

I've worked with some systems that had similar problems, but we had one system, I think it was the one developed by CitySearch, that allowed us to force cache on submission. It worked nearly instantaneously. Nice!

Now, if we can only get news sites to update their search index at the same time! Wouldn't that be a miracle...

Comment by Jay Small, posted November 3, 2002, 8:07 pm

The thing that amazes me is, this is not a difficult matter to overcome, even for a CMS with caching. Smart caching overwrites files to disk immediately when the database entry is changed, then leaves cache files alone for the (many) files that don't change that often.

Even open-source caching, such as that built into the Smarty template engine, supports a scriptable overwrite method that eliminates the problem of caches updating only at intervals. It's silly to trigger cache updates only on a timer.

POST A COMMENT on “Breaking news, delayed by the CMS”

Your name: (optional)

Your site: (optional)

Comment: (only <a>, <em> and <strong> allowed)

Temporarily unavailable news sites

The Web sites of some Gannett newspapers in New Jersey — the Asbury Park Press, the Home News Tribune and the Ocean County Observer — are running this message above their top stories:

Due To Software Improvements, many of the IN JERSEY web sites will be temporarily unavailable after 12:00 pm on Saturday October 12th. The site will be completely restored by Sunday October 13th. We apologize for any inconvenience this may cause

Would their print edition newspapers skip an entire day due to “software improvements”? News sites have a journalistic obligation to be available 24/7/365, even during planned maintenance. The “dead end” sign on the little animated under-construction graphic says it all. I guess anyone at these papers with a clue about the Web is permanently — not just temporarily — unavailable.

Comment by Adrian, posted October 11, 2002, 12:04 am

Nice comment system, my friend, but it looks mighty familiar. :-)

I like the idea of offering <EM> as an allowed tag...I might have to steal that idea to make things even.

POST A COMMENT on “Temporarily unavailable news sites”

Your name: (optional)

Your site: (optional)

Comment: (only <a>, <em> and <strong> allowed)

Knight Ridder URL problems

When Knight Ridder redesigned its newspapers’ Web sites in February, they broke links and gave the sites a unified, restrictive design. J.D. Lasica criticized the “cookie-cutter” look:

It’s as if the corporate designers said, “Miami? I know! Let's throw a palm tree into the masthead.”

But the new Knight Ridder design also allows URLs that make no sense. What newspaper’s articles will you get from links like these?

At the very least, URLs like these should return 404 errors rather than causing reader confusion. Does www.tallahassee.com/mld/philly/news/weather/ give you weather in Florida or Pennsylvania? Does www.dfw.com/mld/pioneerpress/news/editorial/4205885.htm give you the editorial board’s collective opinion at the Fort Worth Star-Telegram or the Saint Paul Pioneer Press? Granted, since it takes mischievous geeks like me to create these funny URLs, ordinary readers might not encounter them ever. But Knight Ridder’s programmers should have anticipated this chance for serious confusion and prevented it.

The other problem with this system is that it makes URLs longer than necessary. Why couldn’t they do URLs like www.philly.com/news/ rather than www.philly.com/mld/philly/news/?

In that second URL, it’s the “philly” after the “mld” that actually determines which of the many Knight Ridder or RealCities sites you see, not the domain name. This is because the dozens of Knight Ridder domain names, like www.thestate.com and www.charlotte.com, all resolve to a single IP address: 66.54.5.40. Rather than taking advantage of HTTP/1.1 “Host” data sent by browsers and using the domain name to determine which Web site the reader wanted, Knight Ridder’s servers require an extra “mld” and then the name of the site over again. They use the HTTP/1.1 Host header for an initial redirect to add that extra “mld” and site name, but afterward the domain name doesn’t matter. Their URLs might as well be written http://66.54.5.40/mld/macon/, because the IP address is as unhelpful as its DNS equivalent.

The domain name ought to be the most important part of any URL. I question why Knight Ridder has made it irrelevant.

Update October 26, 2002, 11:45 pm: Originally I wrote that readers might not encounter and be confused by these funny Knight Ridder URLs much because a “mischievous geek like me” would first have to link to a Knight Ridder site like that. Well, apparently someone has.

Search for “times leader” at Google and the first result is timesleader.com. But on only the second page of Google results, the eleventh result is www.macon.com/mld/timesleader/contact_us/subscriptions_np1/, the twelfth is www.macon.com/mld/thetimesleader/news/, the fourteenth is www.kansas.com/mld/thetimesleader/news/ and the fifteenth is www.kansas.com/mld/timesleader/contact_us/about_np1/. Knight Ridder ought to be concerned about this.

POST A COMMENT on “Knight Ridder URL problems”

Your name: (optional)

Your site: (optional)

Comment: (only <a>, <em> and <strong> allowed)

Timestamps at nytimes.com

About a week ago The New York Times began adding timestamps to individual articles on its front page. Perhaps it is taking a cue from the blogging movement’s style of Web journalism, with frequent updates made obvious by timestamps and blog entries in reverse chronological order.

Since printed newspapers have always presented news in order of importance, most news Web sites do too. For those who read nytimes.com every few hours, the site needed to clarify that a developing story had been updated with more information. But sometimes that new information — called a “writethru” of the story in wire service terminology — does not justify a new headline or new subhead or reordering the stories on the front page. Perhaps the writethru just adds detail, making a 200-word brief into an 800-word report as more information emerges. Regular visitors can now look for any little red timestamps to tell what is new in the past hour, while occasional visitors can read headlines to tell what is new in the past day.

This technique is also used at a bunch of other news sites, among them:

Some other sites use the word “Update” or “New” or even a little graphic symbol to indicate updated stories, but do not use timestamps:

I think the timestamps are an excellent technique for more news sites to adopt when they update a breaking story throughout the day. It certainly is much more enlightening than a general “last updated” timestamp for the entire front page.

On the other hand, this system may give readers incorrect impressions of when the news originally occurred. If that article was last updated at 5:32 p.m., did the event it describes happen in the afternoon? Or the morning? Or last night? There’s no way to tell from these kinds of timestamps when the news occurred or when it was first reported. Posting timestamps like this requires readers to mentally separate the headline and story summary, which is news about an event, from the timestamp, which is “meta-news” about the story itself. A clearer system would be to use both modification and creation dates — both an “updated” time and a “posted” time, which would be closer to when the event occurred. But the two-timestamp system would probably take up too much space on newspaper home pages.

POST A COMMENT on “Timestamps at nytimes.com”

Your name: (optional)

Your site: (optional)

Comment: (only <a>, <em> and <strong> allowed)

Dispatch.com charges non-subscribers

Starting today, the Columbus Dispatch’s Web site costs $4.95 per month if you don’t subscribe to the newspaper’s print edition.

E-Media Tidbits regularly discusses the failures (and very few successes) of paid online content. An important rule of thumb seems to be that few users pay when similar content is free elsewhere. For example, only 1.1 percent of salon.com readers pay for Salon Premium, and only 0.3 percent of WeatherBug’s users pay.

Since TV and radio (and their Web sites) are free, dispatch.com is going to lose online readers — and thus the ability to sell ads to a larger audience. Of course, in the current online advertising slump those page views are not worth much revenue anyway, but the Dispatch’s online strategy won’t help it in the long term if it wants to build a larger audience.

Even if some online readers are willing to fork over their $4.95, the Dispatch’s explanation and instructions are very complicated, especially compared to the Washington Post’s recent quick-and-easy implementation of free site registration. The Dispatch does seem to have its heart in the right place, though, allowing free access for public libraries and schools. It will be interesting to see how it works out for them.

Update October 1, 2002, 5:35 pm: As I was writing this, two Tidbits items reported that The Onion will add “premium” content and that El Mundo, a Spanish newspaper, will charge for print edition content.

POST A COMMENT on “Dispatch.com charges non-subscribers”

Your name: (optional)

Your site: (optional)

Comment: (only <a>, <em> and <strong> allowed)

This page last modified on Wednesday, February 13, 2019 at 2:59 pm