The online newspaper as we know it is turning 10. But despite having had a decade to grow up, many newspapers are still struggling to break their traditional 24-hour news cycle and adapt to a cycleless, continuous-news medium.
Created as a metaphor of the static front-page of the print edition, the homepage can be updated constantly, throughout the day. However, editors do not appear to have reached consensus yet about how often the news should change.
A research project that we started last year at the School of Journalism of the University of Texas at Austin examined how some of the country's largest newspapers are taking advantage of the opportunity to offer a continuous news service in their virtual front pages on the web. It has shown huge disparity among their practices.
In the analysis of 30 newspapers from Editor & Publisher's 2002 list of the 100 largest-circulation newspapers, we found that five of them made virtually no updates during the day, while 13 added only a few breaking news stories. Twelve newspapers were the most dynamic, updating their sites constantly.
In its first phase, our study has recorded changes in the homepages of the top 10, the middle 10 and the bottom 10 newspapers from the list of 100 largest-circulation dailies, during a two-week period (excluding Saturdays and Sundays), between June 23 and July 4, 2003.
A special system created by iMorph Inc. was set to take a snapshot of each homepage every hour, analyze the content and send us by email those pages that were modified since the previous snapshot.
Coders analyzed those pages and recorded 19,974 changes during the two weeks, considering only units of information regarded as news. A change could be a headline, a blurb, a graphic, a photo or just a caption, according to a predefined list of categories. One story could have had more than one change counted, depending on the categories it would fit.
The total of changes in each homepage is an absolute number that might bias toward those newspapers with longer home pages. We have not yet calculated a relative number of changes that takes into account the sizes of the various home pages during the studied period.
Each change went through a typical content analysis process that was enhanced with our focus on measuring the frequency of the updates. To further monitor the trend, we are recording the changes of the same newspapers in a similar period this summer.
Of the group of 12 newspapers with the most changes, eight came from the top 10 circulation papers, three from the middle 10, and one from the bottom 10. Overall, the top 10 newspapers in circulation accounted for most of the changes (56 percent) of the sample, even considering that two of them -- the two tabloids from New York -- were among the less dynamic homepages.
The preliminary findings of this research show that the smaller a paper's circulation, the higher the likelihood for fewer updates in their homepages. The middle-level group had 23.9 percent of the overall changes tracked, and the bottom group had 19.6 percent.
It is important to note that our study is focused on changes that we assume are carried out by the newspapers' online editors. Therefore, we do not count the automated news feeds provided by wire services (whenever it was possible to identify them). We did not count non-news items such as the weather and temperature, stock quotes or tickers, and editorials or columnists.
The 12 newspapers with the most updates during the period studied were led by Long Island Newsday, which had an average of 202.7 changes per day. The others were USA Today (148.2), Houston Chronicle (138.4), Chicago Tribune (133), New York Times (127.5), Los Angeles Times (121.3), Washington Post (115.3), Wall Street Journal (85.9), Hartford Courant (84), (Fort Worth, TX) Star-Telegram (68.7), Allentown Morning Call (64.6) and The Oklahoman (60.2).
The Courant (190,312 circulation), the Star Telegram (218,975), and The Oklahoman (199,581) are the only newspapers in the list of the 12 most dynamic Web sites that represent the middle 10 group. The Morning Call (118,859) is the only one from the bottom 10 group to make the most dynamic list.
The two tabloids from New York City did not pay much attention to their Web sites. The New York Post (590,061 circulation) had the third lowest position in number of changes, modifying the Web page virtually once a day, with a daily average of 14.1 changes; and the Daily News (715,070 circulation) had an average of 42.9 changes, including a few updates during the day.
The least dynamic paper in the study was the (Harrisburg, Pa.) Patriot-News, which is also at the very bottom of the E&P 100 top newspapers list, with a circulation of 101,598. The Patriot News homepage had an average of only 9.7 changes per day, reflecting a typical case of what the industry calls shovelware: it usually updates the home page just once a day, shoveling some content from the print edition.
The other four newspaper Web sites were also basically shovelware: Tampa Tribune with an average of 12 changes once a day (only the paper site was counted, not the news portal TBO.com, where the Tribune partners with a TV station), followed by New York Post (14.1), (Worcester, MA) Telegram & Gazette (18.4) and Buffalo News (24.5). In the case of the latter Web site, it belongs to a newspaper from the middle range of E&P's list, with 223,957 circulation.
As those five sites were static throughout the day during the period studied, 13 other homepages showed only a few changes, usually occasional breaking news, aside from the print edition content that was updated once a day: Columbia State with an average of 58.7 changes daily, Virginian-Pilot (57), The Seattle Times (54.8), Knoxville News-Sentinel (54.6), Wilmington News Journal (44.3), Lexington Herald-Leader (43.6), New York Daily News (42.9), Albuquerque Journal (41.5), San Antonio News-Express (40.9), Louisville Courier-Journal (40.6), St. Paul Pioneer Press (34.8), Spokane Spokesman Review (30.1), and Sarasota Herald-Tribune (25.1).
Our research dug deeper into the results to find how online newspapers updated or changed midday by looking at the 30 newspapers from 11 a.m. to 4 p.m. local time. The leader remained Long Island Newsday with a total of 872 changes over the two weeks sampled, followed by Chicago Tribune (573), USA Today (508), Houston Chronicle (484), New York Times (431), Washington Post (395), Los Angeles Times (334), Wall Street Journal (271), Hartford Courant (254), Star-Telegram (250), Allentown Morning Call (204), and Spokane Spokesman-Review (191).
When we compared the list of the most dynamic newspaper Web sites in the time slot of 11 a.m. to 4 p.m. with the 12 most dynamic newspapers overall, we found just one difference: The Oklahoman, which was in the overall list, was not any more when we checked the changes only between 11 p.m. and 4 p.m. The Spokesman-Review took its place. But the strong similarity of both lists shows that overall the most dynamic papers were consistent throughout the day in how much they focused on changing the homepage during this timeframe.
Recent studies on dayparting have shown that online users tend to visit news websites during the morning and afternoon hours, and these high frequencies show how these online newspapers were providing changed or new content to online users during this key daypart.
We have examined at not only the frequency of updates, but also the nature of such changes. The preliminary results of the two weeks of 2003 have shown that almost half of all changes (47 percent) represented hard news. Thirty percent of changes were soft news, and 22.9 percent were a combination of both (or cases where the coders found it difficult to distinguish).
The predominant changes were in text (91 percent), with very little units of change counting for photos (5 percent) and graphics (0.2 percent). The majority of the changes involved putting new content in the page (72.8 percent), replacing other content. But the content was modified in 13 percent of the cases, its format was altered in 5 percent, related stories were added in 4.7 percent of the sample and additional feature accounted for 4.1 percent.
The content analysis of the two weeks also showed that national news dominated the changes in terms of geography with 43.2 percent, followed by local and regional (39.2 percent), and international (17.6 percent). By category, the leader was business and economy with 16.3 percent, sports (14 percent), followed by politics and government and military/war (with 13.4 percent each), entertainment (9.6 percent), crime (7.1 percent), courts (6.6 percent), and accidents (4.3 percent).
The snapshot of the two weeks showed that most of the newspapers did not add many elements to the stories posted in their Web sites. Only 8.75 percent of the total, or 1,748, changes represented additional elements to news stories. Half of those cases (50.7 percent) were related stories, followed by video (18.3 percent), slide show (9.6 percent), audio (4.6 percent) and maps/charts (1.9 percent).
Another aspect of this project is to analyze the movement of stories within the page, concerning an editorial decision to give a story more or less importance in the site's homepage. Considering that most of the pages analyzed are long, sometimes with several dozen units of information, it was a surprise that only 8 percent of the overall changes represented stories moving up or down the homepage.
It has been intriguing to observe that in a dynamic medium news stories can simply disappear, after occupying for a while the most prominent positions in a newspaper homepage, instead of moving down to secondary places.
By continuing the monitoring and analysis of this sample, we believe we can make a small contribution to the understanding of this genre of journalism, which after a decade of development, is still under construction.