Mar 21 2007

rss has a flavor

Thwwwp Did you know that well over half of the requests seen by del.icio.us are for RSS feeds? That means that people cruising around our site in browsers are actually in the minority, when it comes down to raw traffic. Instead, our heaviest hitters include personalized home pages, desktop news aggregators, and even stranger things.

With that in mind, it makes sense to remember that small changes can have a big impact. Thus, given that our feeds have been doing a decent job this far, they haven’t changed much from their austere beginnings.

Decent isn’t great, though. So, as the team’s semi-official (and published!) feed junkie, I’ve been working on some small improvements. I’ve been looking into how our feeds are used and how to better streamline and present our information in different contexts. These improvements include features like:

  • including tag descriptions in feed titles and descriptions where available;
  • offering the ability to save bookmarks straight from your feed reader;
  • displaying an up-to-date count of saves, without making items appear new again in feed readers;
  • building more useful feed content with links to people, tags, and more bookmark details;
  • providing more metadata where it seems useful, or less where it appears redundant.

And, this is just the start. We’re rolling these changes out gradually, on a per-user-agent basis, and we’re planning for more. So, if you don’t see any improvements in your favorite feed reader yet—or if the changes haven’t quite hit their mark for you—be sure to contact us and let us know!

Les Orchard · deusx Tags: features bookmark this

52 Comments

  • Wally Punsapy  |  Mar 21 2007 at 5:49 pm

    Looks feeder-lickin’ good! Yum! Delicious! No pun …. yeah, yeah.

  • Nick Gerakines  |  Mar 21 2007 at 6:15 pm

    Wally, why are you always the first person to comment on everything?

  • Wally Punsapy  |  Mar 21 2007 at 6:44 pm

    Nick, Joshua tipped me. It’s good to have friends on the inside ;)

  • Joshua Schachter  |  Mar 21 2007 at 9:23 pm

    DO NOT LICK

  • Ian McKellar  |  Mar 21 2007 at 9:49 pm

    It doesn’t look like you’ve turned these on for curl’s user-agent yet :(

  • fakeisthenewreal  |  Mar 21 2007 at 9:49 pm

    Thanks for this! It makes my life better. Well, my del.icio.us-reading life, anyway.

  • cheezburger  |  Mar 21 2007 at 9:59 pm

    plz lv this to profeshionals at ichc!

  • l.m.orchard  |  Mar 21 2007 at 10:39 pm

    @ian: Actually, curl is one of the user-agents for which we probably won’t turn the enhancements on. It’s very generic and hard to be sure what’s actually consuming the feed - so curl will get the most vanilla content.

    However, you might try supplying another user agent like so:

    curl -A “MyNiftyNewsReader” http://del.icio.us/rss/deusx

  • l.m.orchard  |  Mar 21 2007 at 10:42 pm

    …supply a specific user agent, and request it to be whitelisted for enhanced content, that is.

  • Chris  |  Mar 22 2007 at 1:11 am

    many of these rss feed calls are coming from feedmashr.com

  • joshua schachter  |  Mar 22 2007 at 2:29 am

    Maybe we ought to have an optional parameter that would cause the formatting. /rss/whatever?fancy

  • Jan  |  Mar 22 2007 at 5:25 am

    I’m using the rss feeds instead of bookmark folders in my Firefox bookmarks sidebar. For me a basic rss feed transmitting only the link and the description is sufficient. No “bookmark this” etc. in the rss feeds would make them much easier to read.

  • Brett  |  Mar 22 2007 at 9:20 am

    Thunderbird 2.0 (recent nightly build) is getting some odd characters in two of the links in the html:

    -  bookmark this on del.icio.us

    - more about this bookmark…

  • jirkap  |  Mar 22 2007 at 9:45 am

    There are obviously some problems with del.icio.us RSS feeds in Google Reader. New feeds are displayed late or not at all, but in most cases are available under feed’s name (they are just not marked as new)¨… I don’t know if it’s a problem of yours or Google. Just to let you know.

  • Glen Farrelly  |  Mar 22 2007 at 10:47 am

    The feed improvements sound great, but since yesterday I haven’t been able to get any new feeds working in Google Reader or my Blog. Used to work, just doesn’t anymore - except in MyYahoo.

    Wonder if others have this problem and will is it related to these upgrades? Any ETA on fixes?

  • Glen Farrelly  |  Mar 22 2007 at 10:52 am

    The feed improvements sound great, but since yesterday I haven’t been able to get any new feeds working in Google Reader or my Blog. Used to work, just doesn’t anymore - except in MyYahoo.

    Wonder if others have this problem and will is it related to these upgrades? Any ETA on fixes?

  • l.m.orchard  |  Mar 22 2007 at 10:55 am

    @Brett: Hmm, strange. I just checked a few feeds in Thunderbird 1.5.0.10 and Thunderbird 2.0pre (20070322) on OS X, and can’t reproduce the odd characters. For what it’s worth, the one thing that’s unusual about those links, is that there are non-breaking space entities () in the HTML to help keep those links in one piece on end-of-line word wrapping. I suppose it might be worth it to just nix those if there’s an issue, and live with word wrapping in the links.

  • l.m.orchard  |  Mar 22 2007 at 11:04 am

    @Glen Farrelly: The Google Reader thing certainly isn’t intentional, but it seems odd - I’ve got del.icio.us feeds subscribed in my reader there that seem to work fine, but can’t seem to add new one. Looking into it! As for your blog, the odd thing is that that’s one of our JavaScript-based linkrolls and not an RSS feed. That shouldn’t have changed with this release. Looking into that too! As for an ETA - not sure, will need to see just what the issues are first.

  • shawn  |  Mar 22 2007 at 11:07 am

    I just want plain link feeds with my descriptions, how do i get rid of this new crap? Just broke a bunch of stuff. YAY new features.

  • Glen Farrelly  |  Mar 22 2007 at 12:37 pm

    @ l.m.orchard: All my del.icio.us feeds are working now!! Did you fix it?

    What a relief anyway. I love my tag feeds!!

    It was weird as I couldn’t add any new feeds to Google Reader but my old ones worked fine. I could even delete the old ones and add them back in and they worked, but no new feeds would work. They all work now.

    The Java-Script based linkroll had two problems 1) the no new feeds problem I described above and 2) a title must be entered or the whole functionality doesn’t work in Blogger.

    It’s working now though!

    Thanks and I’m looking forward to del.icio.us feeds enhancements, I think they are tremendously useful!

  • l.m.orchard  |  Mar 22 2007 at 1:54 pm

    @Glen Farrelly: Strangely enough, I haven’t actually fixed anything yet :) I’ll keep an eye on your blog sidebar and keep looking into the issue - if it broke once and fixed itself, it might break again.

  • l.m.orchard  |  Mar 22 2007 at 1:56 pm

    @shawn: Can’t please everybody all of the time. :) But, that said, one of the next changes we’re going to make very shortly is to offer an override parameter to force either plain and unadorned feeds or the new enhanced style. Could you tell me what aggregator / feed reader you’re using? We might be able to tailor some things to work better there, along with offering the universal off switch.

  • Yaron  |  Mar 22 2007 at 2:06 pm

    Some of these changes and additions sound really nice and useful, thanks.

    But… I strongly support the suggestion by Joshua. If you want to make changes to the feeds, but only in some cases, then provide a different URI, or a parameter.

    Changing what is returned by the same feed URI based on the user agent is a really really (Did I say “really” already?) bad idea.

    When someone asks for a feed, they should know what it contains and what to expect. And it should look the same for all standard user agents, and contain the same data.

    If two people compare what they get from the same feed, and see they receive different things, an explanation that it’s because they use different readers, etc, etc, won’t really convince them that it’s normal. A feed at a specific address should contain the same things for anyone asking it.

    And if someone uses two different readers, but access the same feed on both of them, they should get the exact same content as well.

    Plus, though IANAL, it may break guidelines of various search engines, of the sort of not to present different content to them and to “regular” people.

    Just give a different address, or a parameter, and be done with it as simple as that. Let people choose which version of the feed they want.

  • tommie  |  Mar 22 2007 at 2:12 pm

    Funny, I found this article from my RSS reader KlipFolio :P

  • Brett  |  Mar 22 2007 at 2:42 pm

    @I.m.orchard: fwiw, viewing source on the message in Thunderbird shows those characters, and not the nbsp entities. Let me try subscribing to some new feeds for individual tags, and trying to reproduce on another machine later today. I’m currently on a Windows XP machine.

  • l.m.orchard  |  Mar 22 2007 at 4:51 pm

    @Yaron: One of the very next tweaks we’ll be rolling out for these new RSS feeds are ?fancy and ?plain parameters that will allow you to override our user-agent detection and force unadorned or enhanced HTML feed content.

  • l.m.orchard  |  Mar 22 2007 at 5:10 pm

    @Yaron: Oh, but with respect for trying to “look the same for all standard user agents” - the problem is, there are no standard user agents when it comes to RSS. There’s a maze of twisty user agents, none alike. The same feed content is rendered differently in various classes and implementations of feed readers. Some feed readers show metadata beyond title / link / description, and some don’t. Some look like email, some are headlines-only boxes in a portal, and some look like rivers-of-news. The pragmatic decision made here is that we’d like to tweak the presentation of our data to account for some of the more popular contexts. It’s a bit quixotic, but FeedBurner, for example, has been pretty successful at satisfying demand for this sort of thing.

  • wet blanket  |  Mar 23 2007 at 1:36 am

    So… the majority of hits are passive, i.e. - when someone is reading someone’s blog or firing up their aggregator and checking all their feeds…

    This is like getting excited about web crawlers visting your website.

  • shadow  |  Mar 23 2007 at 2:10 am

    The feed improvements sound great, but since yesterday I haven’t been able to get any new feeds working in Google Reader or my Blog .

    the same to me, why?

  • Martin Kelley  |  Mar 23 2007 at 4:14 am

    Oh come on, you really didn’t throw that tag information in the Description field, did you?!?! A lot of us have elaborate sand castles (http://www.quakerquaker.org) built up our Del.icio.us RSS feeds. I’ve trained my whole network to use it and I don’t want unexpected changes like this to wreck what I’ve had. You guys keep us with trust.

    I’m glad you’re innovating (metadata on the categories is cool) but I third the motion that you need to roll these out in such a way that it doesn’t change the existing feed. I especially don’t like Del.icio.us advertisements in my Description field. I’m a totally publicity whore for you guys but I want to do it on my terms. Now my pages have unwanted gibberish, some of which exposes things I don’t want exposed. Not cool.

    Pleeeease turn this off and turn it back on with an extra switch of some sort. I really want to know you guys are responsive to the needs of those of us pushing both Del.icio.us’s boundaries (you should see the elaborate Yahoo Pipe I’ve built on top of spliced Del.icio.us RSS feeds) and your user-base by signing up our friends.

    Martin Kelley, http://martinkelley.com/

  • Johan Liesén  |  Mar 23 2007 at 7:45 am

    Will you guys prefix the resource-attribute inside the topics-bag with “rdf:” sometime soon? I was nagging about this about 8-9 months ago, but I guess it ended up at the end of the todo list.

  • l.m.orchard  |  Mar 23 2007 at 10:52 am

    @Martin Kelley: One of the next changes we’ll be rolling out are ?plain and ?fancy switches to override our user-agent detection. However, that detection should only supply the new enhancements to user-agents that we think can use them. Have these changes broken any of your RSS feed readers? If so, let us know, and we can make sure that the unadorned stuff gets fed to those. Generic RSS parsers like Magpie, Rome, feedparser, etc should not be seeing the new HTML content. On the other hand, we also have linkrolls and JSON feeds containing much of the same data, as an alternative.

  • l.m.orchard  |  Mar 23 2007 at 10:53 am

    @Martin Kelley: Although, I do see that the new content is showing up in Pipes, which was not what we intended. So, that should be another fix in the next release.

  • Paul Irish  |  Mar 23 2007 at 10:56 am

    I’ll throw my voice into the pot of people wanting the old version. When I’m dealing with >100 new posts in my delicious popular feed, its a bit annoying to have all these other links and cruft. I’m looking forward to an override to simplify the feeds.

  • Chris  |  Mar 23 2007 at 12:23 pm

    The blue numbers are quite large and loud in Bloglines.

  • JS  |  Mar 23 2007 at 2:09 pm

    RSS all over..

  • Maritn Kelley  |  Mar 23 2007 at 6:14 pm

    @l.m.orchard: thanks for the feedback. You’re right that the new stuff doesn’t show up with Del.icio.us read directly through Magpie. Where it’s gone public on me has been with feeds pulled together by Yahoo Pipes. So I’ll be a happy camper again when that’s fixed (and I’m happy knowing you plan to fix it). Thanks again.

  • Xochitl  |  Mar 24 2007 at 2:22 am

    the badges, sometime they are very long. It will be nice to have them in a loop, so the actual seen space is shorter

  • Richard Barta  |  Mar 24 2007 at 1:54 pm

    Is it possible to get specific search results in an RSS newsfeed? If so, how? If not, why not?

  • Alexandre  |  Mar 24 2007 at 3:51 pm

    Very Good you Works..

    I like you Blog (and Site)…

    I’m from Brasil…

  • scottlord  |  Mar 25 2007 at 10:31 pm

    My del.icio.us tage button and toolbar button hasn’t been working for a day or two; could you write me.

  • Michael Massing  |  Mar 26 2007 at 6:37 am

    This may be somewhat off-topic, but I thought I’d ask in a general way here before trying the support/enhancement request route. Google’s Webmaster Tools will accept an RSS as a form of sitemap, but they say that del.icio.us prohibits their access via robots.txt at the domain level. I’d like to be able to allow this, since as my link shows, I use one of my delicious bundles as a channel for a sort of light-footprint, low-maintenance blog. Given that you’ve been looking at the varied uses people (and programs) make of your RSS traffic, is allowing spiders access for indexing purposes (possibly switchable by user) something that might ever happen? Thanks.

  • Ben  |  Mar 26 2007 at 8:27 am

    I strongly agree with Joshua and Yaron. Serving different format/content based on user-agent is a really bad precedent for feeds. RSS should be really simple syndication, not gold/silver/bronze, where the desired visitors get extra content. Do that with web pages, do that with your JSON feeds if you want. But RSS? That’s just a mess.

    Simple example: delicious RSS feed X gets spidered by some feed directory site, and described as having various attributes which are available to the user-agent associated with that spider. Along comes Joe User, browses the site, reads the blurb about feed X, adds it to his feedreader. But supposing Joe’s feedreader isn’t among the friends-of-delicious user-agents. Then Joe will find the feed with a reduced and/or different set of attributes. Sure we’re just talking about “extras”. The title, link, and description will presumably be the same across all “flavors”. But this context/flavor switching just invites confusion.

    It would be much more honest, easy to understand, intuitive, RESTful and all those good things to just sort by URL. And don’t make the base URL dynamic, so users have to add ?plain if they want to load something which behaves consistently across all user-agents. Rather it should be the other away around; if you must offer a user-agent context switching format (though I don’t see why), use a ?flavor flag which (optionally, secretly in a cryptic no public list of recognized useragents way) adds extra attributes.

  • macdet  |  Mar 26 2007 at 11:21 am

    i cant live without del.icio.us :)

    Now i have to bookmark a lot of acww tags too help my little daughter!

    Plese help to build a nice info page!

    http://wiki.mobbing-gegner.de/Clarissa_Rieger/Animal-Crossing-Wild-World

  • simon  |  Mar 26 2007 at 2:41 pm

    great stuff - del.icio.us & bloglines have made my life even easier…

  • Marios Lublinski  |  Mar 27 2007 at 12:26 am

    Great Job del.icio.us keep up the good work.

  • kl  |  Mar 27 2007 at 6:34 am

    Go for Atom 1.0.

  • Brian  |  Mar 27 2007 at 5:18 pm

    Nooooooooooooooooooooo!!! Please implement an optional switch for the bookmark this and tags in the description fields. You have taken a very clean look for our page and made it so horrible it is unusable. None of our website visitors will use this. It is like mud on a windshield.

  • dameer  |  Mar 28 2007 at 4:34 am

    Are these improvements running or else? I don’t see any changez like “displaying an up-to-date count of saves, without making items appear new again in feed readers”…

  • l.m.orchard  |  Mar 28 2007 at 8:00 pm

    In advance of getting another blog post out and updating help pages, I’ll note some of today’s launched fixes here: You can now add ?plain onto the end of any RSS feed URL to disable the new HTML content, and add ?fancy to turn them on. In addition, we’ve taken some user agents out of the sniffer, which should stop the new content getting to Yahoo! Pipes and a few other applications (such as Hexamail’s News2Web).

  • dameer  |  Mar 29 2007 at 1:35 am

    …yeah, great job, keep movin’!

  • Mountain/\Ash  |  Mar 29 2007 at 8:26 am

    Is the feed still limited to the last 100 entries? Can their be a query string request to get the lot?

Recent comments

Feeds

Who are we?

Posts by tag

Posts by month