杏吧原创

AT SEVEN o鈥檆lock every evening, Robert Scoble settles down on his sofa with his tablet PC and checks what鈥檚 new on more than a thousand websites. It takes him about an hour. When he鈥檚 through, he will post a commentary on what he has read on his weblog, or blog. Within minutes of being posted, his comments will be found and read by thousands of people.

But Scoble does not have to click laboriously through a thousand bookmarks. And readers of his blog won鈥檛 need to spend their evenings hitting 鈥渞eload鈥� just in case he posts something new.

Instead, Scoble and his readers use news harvesting software that shows them which of their favourite websites has posted a new article. So among Scoble鈥檚 thousand sources, for example, he鈥檒l only look at the hundred or so with something new to say.

RSS, short for Really Simple Syndication, is rapidly becoming the most popular way for websites to notify their readers that a new article has been posted. RSS comes in two parts 鈥� a feed and a newsreader. A website continuously scans itself for any new postings and then automatically updates a small file, called an RSS feed, which can be picked up by a newsreader. At newscientist.com, for instance, our RSS feed is updated with breaking news headlines.

The newsreader installed on a user鈥檚 PC checks the RSS file for fresh postings at intervals chosen by the user. If a new item is found, its headline pops up on the user鈥檚 screen in a browser-like interface, which lists all the feeds the user has chosen to monitor. In this way, the software gives users near-instant updates from as many websites as they chose.

This means that the days of having to keep visiting websites 鈥� or subscribing to countless email bulletins 鈥� in search of news, are over. The range of RSS news subjects is breathtaking. Technorati.com, a search engine for RSS feeds, now sifts through more than 2 million sites offering feeds. And newsreaders are available not only for PCs but also for internet-capable mobile phones, PDAs and even the Apple iPod.

But not everyone is completely happy with the boom in RSS. Werner Vogels, a distributed systems researcher at Cornell University in Ithaca, New York, fears that ever more newsreader packages automatically asking websites 鈥渨hat鈥檚 new?鈥� will lead to an increase in connection costs for website owners. He says without urgent fixes, the bandwidth consumed by RSS will impose a significant burden on popular sites and force them to expand their connection capacity.

The problem, says Vogels, is that Really Simple Syndication is a little too simple. The RSS feed summarises the changes as a list of latest articles, which may consist of a title, a description and a web link. Most RSS files contain the last day or so鈥檚 new items, but there is no limit on how big the file can be. With frequent updates, and full descriptions of their content, some can grow very large, often to over a megabyte.

And every time that file changes, every RSS newsreader visiting the site will download it, hogging connection capacity. And some newsreaders will do that as often as once every minute. So popular sites risk RSS overload.

Busy news sites like Slashdot.org have learned to keep a close eye on RSS activity. 鈥淚n terms of page requests, RSS is roughly 20 to 25 per cent of our total,鈥� says Rob Malda, the site鈥檚 founder. 鈥淗owever, we are extremely restrictive on RSS. We only update the RSS feeds every 30 minutes. Anyone loading them much more than that will get their machine locked out.鈥�

Happily, the answers to RSS鈥檚 problems may also lie in its simple nature. Fixes that help websites keep their load down also work with RSS. For instance, a web browser generally sends a 鈥渢ag鈥� to the web server, stating the time it last visited. If web pages have not been updated since that time, the server sends nothing, telling the browser to use the old copy saved (cached) on the PC.

RSS readers send the same tag, so suppliers of RSS feeds could use them in a similar way. Instead of sending the whole file, they could send only the items updated since the last check. Vogels suggests that this simple fix would cut RSS-related traffic by 50 to 60 per cent. But he believes that the real solution would be far more complex. 鈥淚t will take a major internet news overload 鈥� such as another 9/11 鈥� to encourage people to switch to a more distributed protocol,鈥� he says. Such a protocol might quell bandwidth demand by working like a peer-to-peer file-sharing application: newsreader could share the RSS feeds they had collected with others.

Others are less concerned. Cory Doctorow, an editor with popular culture website boingboing.net, says: 鈥淩SS鈥檚 simplicity is critical to the way ideas fan out over the internet. Bandwidth is a small price to pay.鈥�