intertwingly

Hell, sometimes I comment just to say "look at me." Um, like right now.

So let me get this straight: a guy with a weblog reads a post on a weblog about automated weblog post excerpts, and immediately posts an item on his weblog that links to the weblog post about automated weblog post excerpts, fully aware that this will cause his weblog post to be automatically excerpted and included in the list of weblog post excerpts under the weblog post about automated weblog post excerpts... and then he complains about "look at me" self-publication?

Hey Mark,
I now this is presumptious of me to ask this but could you go back to adding the referrer hit count to your linkbacks instead of or in addition to the excerpts? I found them more informative, more pertinent and less noisy than the excerpts.

Dare: read the second paragraph for instructions on how to see the referer counts.

I didn't get the impression that he was saying that "look at me" is bad, so much as that having to click your own link, knowing as you do that it will disappear in a storm of links from bigger names, was bad. Those strike me as the best two things about pingback: every ping is equal, and it removes some of the awkward social nuances of linking and having your link noticed.

And, Dare, the counts are still there, in the title attribute of the link, but how on earth are they more informative and pertinent? If Dave Winer says "Mark Pilgrim: thousands die." and 300 people follow the link, and Joe Blogger says "There are several things that Mark missed in his analysis of the first deadly computer virus, including one that may well save your life." but only three people follow the link, which would you rather read?

I say, axe the counts and the sort by count, too: I already know that Dave and Zeldman are more popular than kryogenix and I. Sorting referrers in reverse-chronological order makes the same sort of sense as it does for posts. Would you change the order of your posts on the front page to list them from most popular to least?

Here's a dirty little secret: the sort order is accidental, a byproduct of other unrelated logic of the larger script of which the linkbacks are a small part. That said, it would be nontrivial to change it to reverse-chronological-by-first-clickthrough, since nowhere in the larger script do I save or deal with dates in any way.

I'm incredibly busy this weekend, but I'll see what I can do, maybe next week.

Phil,
I rarely track who's who in blogging circles besides the people on Sam's blogroll. When I want to "read more" about Mark's links I typically click on links to people who I'm familiar with and barring that aim for popular links since it is more likely I'll find good opinions and comments at the end of such links instead of MLP (Mindless Link Propagation - a K5ism).

I realize this discriminates against low traffic blogs with quality opinions but that's a trade off I'm willing to make to save my browsing time.

Another way to look at Dare's recent comment: referer counts provide an indirect measure of relevance much in the same way that google does. Should google sort by date? I think not.

This being said, the implementation I'm contemplating will be the product of a "dirty little secret" itself. Like I have done with trackbacks and pingbacks to date, linkbacks are likely to appear as comments - and therefore in cronological order (or reverse cronological order in comments.rss).

Sam: new discovery, you have two permalinks for each post. I did not realise this; from now on, I shall try and ping the "real" archive from the Comments link rather than the daily archive from the "permalink" link.

Mark: referrer entries don't really have to be sorted in reverse-chronological, do they, just /not/ in descending order of visitors? Since the list is unordered, in essence, you could do it alphabetically or something, since reverse-chronological then leads to a Slashdotesque First Post (well, First Link) mentality? It must worry you slightly that you're now popular enough that there's a possibility that people will play those sorts of games to get a better mention...

Phil: our day will come. :-)

Dare: I think that this is why having page excerpts attached to your linkback is good; Simon Willison does it for pingbacks, and Mark does it for referrer-log links.

Pingback from as days pass by: Referrer links

'[I]t seems to me that one could view pingback as nothing more than an automated "look at me"'
I admit that you're right here, Sam. The thing here is twofold: first, that it doesn't *feel* like shameless self-publication if the system does it *for* you, and secondly, if you are right, then all forms of attempting to generate links that go "back up" the link chain (whether *Back, putting hits in referrer logs, webrings, whatever) are like that, and therefore the system should do it automatically because then you don't have to do it manually every go. (This is the "people will take drugs anyway, so let's make them *safe* drugs" argument, I think, so your reaction to it may depend on your opinion of that.)

Should Google sort by number of hits, making Yahoo the number one result for every single search?

Actually, the equivalent of BlogPageRank wouldn't be a bad idea: starting from Phil Pearson's data you could make a pretty reasonable BlogPageRank. Of course, the Big D and the Big Z will still end up on top, but at least it would be less discriminatory toward people who are mostly read as RSS, and deliver the bulk of their referrals from 127.0.0.1:5335.

Actually, all other things being equal, if you are looking for "further reading on today's posts," you want the links that are delivering the fewest visitors, not the most. My experience with my own links to people who show referral counts has been that if I do my usual, going on and on explaining and describing and cautioning and overthinking and explaining some more, I deliver about a quarter as many hits as when I do an old-school "Mark has a cool new referral script" link.

Ah, and wandering back to the original topic of your pingability: for some reason, my MT 2.5 bookmarklet isn't picking up your TrackBack RDF. Ben did change the spec to make it better RDF, but it's supposed to fallback to the 2.2 style, and I can't see why

my $url_no_anchor = $url) =~ s/#.*$//;
my(@items);
while ($c =~ m!(<rdf:RDF.*?</rdf:RDF>)!sg) {
my $rdf = $1;
my($perm_url) = $rdf =~ m!dc:identifier="([^"]+)"!;
next unless $find_all ||
$perm_url eq $url || $perm_url eq $url_no_anchor;
## We look for trackback:ping, but fall back to about
## (used in MT 2.2).
my($ping_url);
if ($rdf =~ m!trackback:ping="([^"]+)"!) {
$ping_url = $1;
} elsif ($rdf =~ m!about="([^"]+)"!) {
$ping_url = $1;
}
my($title) = $rdf =~ m!dc:title="([^"]+)"!;
if (!$title && $rdf =~ m!dc:description="([^"]+)"!) {
$title = MT::Util::first_n_words($1, 5) . '...';
}
push @items, {
ping_url => $ping_url,
permalink => $perm_url,
title => $title,
};

wouldn't be picking up your RDF.

Phil - I'm confused. When I run that code (with an addition of an open paren on the first line and a close curly brace on the last line) on this page, I get the following results:

ping_url => http://www.intertwingly.net/blog/905.rdf
permalink => http://www.intertwingly.net/blog/905.html
title => Still pingbackable

Furthermore, I do see evidence that you have sent me a successful trackback. How was this trackback done?


Posted by Sam Ruby

I cheated: rather than do it from the bookmarklet, I just grabbed the url-to-ping from the page source and pasted it into the textarea in the main MT posting page. I was in a hurry and didn't want to try to figure out what was going on with the bookmarklet.

And with good reason: an hour and a half later, I found the bug where the program gets back a single trackback url from that sub and set the param "to_pings_urls" rather than "to_ping_urls", so the template for the bookmarklet didn't find anything for "to_ping_urls". Sorry to have dragged you into it - I didn't notice until well into the debugging just now that I also couldn't find the TrackBack url on my own individual archive pages.