Comments for DeWitt Clinton http://blog.unto.net indeterminate things Sun, 07 Mar 2010 21:10:44 +0000 hourly 1 http://wordpress.org/?v=3.3-aortic-dissection Comment on Blogging on Buzz by Danny Tuppeny http://blog.unto.net/blogging-on-buzz/comment-page-1#comment-3818 Danny Tuppeny Sun, 07 Mar 2010 21:10:44 +0000 http://blog.unto.net/?p=800#comment-3818 <p>Is there any way to follow someone using a service other than Google Buzz, on Buzz?</p> <p>I see lots of talk about it being very open, but I've not seen any way to follow people from other services in Buzz. I would love to stop opening Twitter because I can see it all in Buzz.</p> Is there any way to follow someone using a service other than Google Buzz, on Buzz?

I see lots of talk about it being very open, but I’ve not seen any way to follow people from other services in Buzz. I would love to stop opening Twitter because I can see it all in Buzz.

]]> Comment on Blogging on Buzz by DeWitt Clinton http://blog.unto.net/blogging-on-buzz/comment-page-1#comment-3817 DeWitt Clinton Fri, 05 Mar 2010 03:47:12 +0000 http://blog.unto.net/?p=800#comment-3817 <p>No, not yet, but we definitely will. I too find that the various sources have different signal/noise ratios, and would like to slice them up in the feeds myself.</p> No, not yet, but we definitely will. I too find that the various sources have different signal/noise ratios, and would like to slice them up in the feeds myself.

]]> Comment on Blogging on Buzz by Aristotle Pagaltzis http://blog.unto.net/blogging-on-buzz/comment-page-1#comment-3816 Aristotle Pagaltzis Fri, 05 Mar 2010 02:53:25 +0000 http://blog.unto.net/?p=800#comment-3816 <p>Does Buzz offer any feed of just someone’s Buzz posts, excluding their Twitter and blog traffic and all the other noise? (I either follow those directly, or else I don’t care about them.)</p> Does Buzz offer any feed of just someone’s Buzz posts, excluding their Twitter and blog traffic and all the other noise? (I either follow those directly, or else I don’t care about them.)

]]> Comment on A Survey of Rel Values on the Web by Michael Fagan http://blog.unto.net/a-survey-of-rel-values-on-the-web/comment-page-1#comment-3772 Michael Fagan Tue, 17 Feb 2009 19:29:15 +0000 http://blog.unto.net/?p=734#comment-3772 <p>well, you could make several different ranked lists</p> <ul> <li>rel values sorted by total usage</li> <li>rel values sorted by number of web pages they appear on</li> <li>rel values sorted by number of websites (domains?) they appear on</li> </ul> <p>as well as lists of what rel values appear most together (on the same page and/or within individual rel=""s)</p> <p>this could also be used to detect common errors (capitalization, spelling, wrong separators) which could be used in the w3c's html validator or html tidy</p> well, you could make several different ranked lists

  • rel values sorted by total usage
  • rel values sorted by number of web pages they appear on
  • rel values sorted by number of websites (domains?) they appear on

as well as lists of what rel values appear most together (on the same page and/or within individual rel=”"s)

this could also be used to detect common errors (capitalization, spelling, wrong separators) which could be used in the w3c’s html validator or html tidy

]]> Comment on A Survey of Rel Values on the Web by DeWitt Clinton http://blog.unto.net/a-survey-of-rel-values-on-the-web/comment-page-1#comment-3771 DeWitt Clinton Tue, 17 Feb 2009 14:05:47 +0000 http://blog.unto.net/?p=734#comment-3771 <p>Philip - that's a great point about a handful of pages with tons of rel values biasing the results. I definitely saw patterns like "track:track..." that show up further down the list; that's one of the reasons I only published the top 25 tags and cherry picked from there. </p> <p>In this case I really was curious about total usage, as I was wondering specifically about values like 'me' and 'muse', which legitimately appear more than once.</p> <p>That said, I might re-run the test counting only once per page, just to compare. </p> <p>Also, I didn't normalize to lowercase, but in retrospect I should have.</p> <p>Thanks for the link, Philip.</p> <p>Cheers,</p> <p>-DeWitt</p> Philip – that’s a great point about a handful of pages with tons of rel values biasing the results. I definitely saw patterns like “track:track…” that show up further down the list; that’s one of the reasons I only published the top 25 tags and cherry picked from there.

In this case I really was curious about total usage, as I was wondering specifically about values like ‘me’ and ‘muse’, which legitimately appear more than once.

That said, I might re-run the test counting only once per page, just to compare.

Also, I didn’t normalize to lowercase, but in retrospect I should have.

Thanks for the link, Philip.

Cheers,

-DeWitt

]]> Comment on A Survey of Rel Values on the Web by Philip Taylor http://blog.unto.net/a-survey-of-rel-values-on-the-web/comment-page-1#comment-3770 Philip Taylor Tue, 17 Feb 2009 10:18:58 +0000 http://blog.unto.net/?p=734#comment-3770 <p>For comparison, I have <a href="http://philip.html5.org/data/link-rel-rev.txt" rel="nofollow">some rel/rev data from a year ago</a> based on 130K pages from <a href="http://www.dmoz.org/" rel="nofollow">dmoz.org</a>. I counted number of pages rather than total number of occurrences, because that reduces the noise in the data caused by a few pages with huge numbers of repeated values. E.g. <a href="http://l0t3k.org/security/docs/honeypotting/en/" rel="nofollow">this page</a> has a load of rel="chapter", and it's used 12 times per page on average in my data, which explains part of the difference in ranking compared to your data.</p> <p>(Also I didn't split on tokens or convert to lowercase, since I wanted to see the original values.)</p> <p>My data also shows a lot of similar rel patterns like rel="track:track_pagetag=..." and rel="balloon29" and rel="lightbox[pkguitars]", presumably all coming from a handful of pages that are abusing the semantics of rel. It'd be interesting to know how many of your 1.8M unique values are instances of this kind of pattern coming from a small number of pages, and how many are real legitimate-looking values.</p> For comparison, I have some rel/rev data from a year ago based on 130K pages from dmoz.org. I counted number of pages rather than total number of occurrences, because that reduces the noise in the data caused by a few pages with huge numbers of repeated values. E.g. this page has a load of rel=”chapter”, and it’s used 12 times per page on average in my data, which explains part of the difference in ranking compared to your data.

(Also I didn’t split on tokens or convert to lowercase, since I wanted to see the original values.)

My data also shows a lot of similar rel patterns like rel=”track:track_pagetag=…” and rel=”balloon29″ and rel=”lightbox[pkguitars]“, presumably all coming from a handful of pages that are abusing the semantics of rel. It’d be interesting to know how many of your 1.8M unique values are instances of this kind of pattern coming from a small number of pages, and how many are real legitimate-looking values.

]]> Comment on A Survey of Rel Values on the Web by DeWitt Clinton http://blog.unto.net/a-survey-of-rel-values-on-the-web/comment-page-1#comment-3769 DeWitt Clinton Tue, 17 Feb 2009 06:28:40 +0000 http://blog.unto.net/?p=734#comment-3769 <p>Kevin -- Spinn3r crawls content exclusively from feeds, right, or do you have a web crawler going as well? I feel like I've seen hits from your spider in my web logs but I could be mistaken.</p> <p>Either way, it will be interesting to see what values are popular on feed-oriented sites as opposed to the general web. I suspect that certain values will appear just as frequently, such as 'nofollow'. But others, like the 'openid' values, won't be used much at all.</p> Kevin — Spinn3r crawls content exclusively from feeds, right, or do you have a web crawler going as well? I feel like I’ve seen hits from your spider in my web logs but I could be mistaken.

Either way, it will be interesting to see what values are popular on feed-oriented sites as opposed to the general web. I suspect that certain values will appear just as frequently, such as ‘nofollow’. But others, like the ‘openid’ values, won’t be used much at all.

]]> Comment on A Survey of Rel Values on the Web by Kevin Burton http://blog.unto.net/a-survey-of-rel-values-on-the-web/comment-page-1#comment-3768 Kevin Burton Tue, 17 Feb 2009 06:07:30 +0000 http://blog.unto.net/?p=734#comment-3768 <p>Hey..... Thanks for posting this.</p> <p>Yeah. I was thinking about rel="canonical" as well.</p> <p>I suspect this will shoot right near the top.. </p> <p>We're going to be posting automated stats on our crawler in our next release.... which should be any day now :)</p> Hey….. Thanks for posting this.

Yeah. I was thinking about rel=”canonical” as well.

I suspect this will shoot right near the top..

We’re going to be posting automated stats on our crawler in our next release…. which should be any day now :)

]]> Comment on A Survey of Rel Values on the Web by possible248 http://blog.unto.net/a-survey-of-rel-values-on-the-web/comment-page-1#comment-3767 possible248 Tue, 17 Feb 2009 03:50:11 +0000 http://blog.unto.net/?p=734#comment-3767 <p>Oh dear, sorry for the double post, but it appears that my usage of <a> and <link> messed up the comment. I wasn't expecting the HTML to go unescaped.</p> <p>For clarification, that sentence should read:</p> <p>I never really thought about the usage of the "rel" attributes to <a> and <link> and until now. I never understood the point of XFN, and am pretty sure that I’ve only ever used "rel='nofollow'" in my history of web design and web app programming.</p> Oh dear, sorry for the double post, but it appears that my usage of <a> and <link> messed up the comment. I wasn’t expecting the HTML to go unescaped.

For clarification, that sentence should read:

I never really thought about the usage of the “rel” attributes to <a> and <link> and until now. I never understood the point of XFN, and am pretty sure that I’ve only ever used “rel=’nofollow’” in my history of web design and web app programming.

]]> Comment on A Survey of Rel Values on the Web by possible248 http://blog.unto.net/a-survey-of-rel-values-on-the-web/comment-page-1#comment-3766 possible248 Tue, 17 Feb 2009 03:48:12 +0000 http://blog.unto.net/?p=734#comment-3766 <p>Excellent blog post, man.</p> <p>I never really thought about the usage of the "rel" attributes to <a> and until now. I never understood the point of XFN, and am pretty sure that I've only ever used "rel='nofollow'" in my history of web design and web app programming.</a></p> <p>I've known that there were a few people that put more thought of what goes into the "rel" attribute, but your quantification of the usage, and how the attribute is exactly being used, really puts it into perspective.</p> <p>I'd love to see a followup post that explained more about how you got these results. :)</p> Excellent blog post, man.

I never really thought about the usage of the “rel” attributes to and until now. I never understood the point of XFN, and am pretty sure that I’ve only ever used “rel=’nofollow’” in my history of web design and web app programming.

I’ve known that there were a few people that put more thought of what goes into the “rel” attribute, but your quantification of the usage, and how the attribute is exactly being used, really puts it into perspective.

I’d love to see a followup post that explained more about how you got these results. :)

]]>