.comment-link {margin-left:.6em;}

Friday, October 01, 2004

Using TrackBacks for trackacrosses

This is a rather long post that in the end comes to a pretty simple conclusion. TrackBacks are sometimes used to link to posts that one has not actually commented on, but that discuss the same subject as one's own post. In these cases, etiquitte calls for posting an update that "x, y, and z are also talking about this subject."

This raises the question of whether a distinction should be made between TrackBacks from articles that actually comment on the post they link to and TrackBacks from articles that merely cross-reference the linked to post. Is that distinction useful enough to users to warrant a separate button? If so, it would be easy enough to accomplish. HaloScan, MT, etcetera could just add a second window to its "manage TrackBacks" function, allowing individual TrackBacks to be designated either as TrackBacks or TrackAcrosses.

The discussion below comes to this simple assessment after a lot of technical discussion about informational efficiency. There is also a discussion of the relation between informational efficiency and reciprocity as moral standards. If that kind of stuff interests you, read on.

The Black Republican has posted a policy about deleting trackbacks that don't actually link to his posts.

He can do what he wants, of course, but I think this is the wrong response.

I have several times used TrackBacks to link to sites that are commenting on the same thing I am commenting on, without placing an update that links to these other commenters. Why? Because I am not commenting on their posts. I am just providing a link to my related discussion.

This seems a legitimate and important use of TrackBacks, unless someone wants to set up a separate TrackAcross function. One might think it is a simple enough matter update a post with links to any page where one has put TrackBacks, but this labor is not insignificant. It is probably about as much work as posting the TrackBacks. i.e. It doubles the work. If the information value of the link to those looking for information is the same in both directions, that would imply that people who are following a norm of reciprocation should provide both links. (Reciprocation is Black's rationale for his policy: "Trackbacks are kind a of mutual admiration society - each side links to the other. If it isn't mutual, I reserve the right to delete your ping.") But both links are not equally valuable.

TrackBacks are mainly used by low-traffic bloggers, trying let people on higher traffic sites know that they exist. If I put on my post an update with a link to a Captain's Quarters post where I placed a TrackBack, that update link might be as valuable to the individual user who sees it as the trackback on Cap's post is to the individual user who sees that, but a thousand times more people will see the TrackBack link on Cap's site than will see the update link on mine. If I am trying to be efficient with my time in terms of information value for the blogosphere (which is the only reason I blog in the first place) I won't bother with the update unless I think the TrackBacked-from post is especially important.

Indeed, if I think the track-backed from post is crap, I wouldn't provide an update link even if doing so was costless, because that would only misdirect my readers. I only want to link people to what I think is important. If I find a post that is saying crap about something that I have the right answer to, the informationally rational thing is TrackBack without updating. By the same standard (informational efficiency), what definitely should be deleted (if one has the time) is non-related TrackBacks, which simply muck up the information flow for people trying to find related items.

I posted a more general comment a while back (I think it was on Patterico) about the inefficiencies that are created by high-traffic/low-traffic assymetries and how they can be analyzed. I kept a word copy:

I wish all the nerve-center blogs would give a certain priority to posting links to important stuff posted on the periphery. Those of us who are consumers of the new media check a variety of blogs on a regular basis. I don't need links to Hugh Hewitt and Powerline (though I am glad to see them) because I check them out anyway. What is more important for the high-traffic blogs to do is be alert to those who are trying to forward important information, and help direct some eyeballs to what ought to recieve substantial attention, but is likely to remain unknown if higher traffic sights don't take notice.

The new media's success is precisely because of its inherent advantages in this area. Attention can be easily forwarded in networked fashion. The key is to do this efficiently. Eyeballs (and the brain power behind them) are the resources that the new-media seeks to distribute to their most valued use. Eyeball time does two things. It provides information the owner of the eyeballs, and it allows the owner of the eyeballs to forward information to other eyeballs. The former is a constant marginal value. If Michelle Malkin has something important to relate, it is as important to the millionth reader as to the first. In contrast, the forwarding function of eyeballs has diminishing marginal value. If the first hundred eyeballs confer a 50% chance that important information will filter up and receive its proper attention then the second hundred eyeballs will only confer about an additional 25% chance.

For the new media to be efficient, those who are directing eyeballs need to make sure they are accounting the diminishing marginal discovery value of eyeballs. That means that for two items of equal importance, priority should be given to what is coming from less trafficked sources. Information does not just filter up on its own. It needs to be grabbed and pulled up, and it may help if practitioners are self-conscious about accounting both inherent value and discovery value.

TrackBacks are an important tool that low traffic sites can use to do their part to try to direct attention to their own important contributions. To make this tool as inexpensive to use as possible, so that low traffic sites can get the most out of it, the flagging of important information should not be allowed only on condition that they do extra work that is not informationally efficient. Thus my policy on my own site will be to welcome trackacrosses from anybody who wants to use them, but please, if you think my discussion is particularly worthwhile, do link to it.

Lastly, maybe HaloScan and other blog tools should add the functionality to name a ping a TrackAcross instead of a TrackBack. All it would take is the addition of a second ping window to their "Manage TrackBacks" function. After all, this distinction is important information for people who are looking through TrackBack links. It tells them whether a particular link comments on the piece one has just read or whether it is simply a link to another discussion on the same subject.


In response to the comments I left on his post, Chris at Black Republican, left some interesting replies. For one, he added more about the importance of reciprocity (this was without seeing what I wrote about it in the post above):

...The fact is, Trackback was created for congenial reciprocity; using it without reciprocity is nothing but spam, and it's fairly rude.

If you need something more substantial than courtesy as a reason, think of it this way: your Trackback is essentially giving your blog free advertising on my bandwidth, and I'm getting nothing in return. Why should I agree to that, especially if I've never heard of you?

If you think your trackacross idea has merit, go talk to the guys at Slashdot or somewhere else to work out the code. Good luck getting bloggers to agree to include an ideal spamming tool on their sites.

Thinking again about the criterion of reciprocity, I shouldn't say that informational efficiency is a higher principle. Rather, my point is that the two are compatible. The kind of reciprocity I want from others is for them to link to my contributions according to the same principle of informational efficiency I use in linking to others. I'll comment on what I think is worth commenting on and link to what I think is relevant reason or evidence.

That wouldn't work very well if everyone was contrarian and only thought their own thoughts were worthwhile. Then people would just try to direct people to themselves and not link to others except out of self-interest, adhering to a tit-for-tat strategy of reciprocal cooperation to induce people to trade links with them. (Chris's TrackBack policy is a little firmer than tit-for-tat, threatening permanent non-cooperation.) But bloggers are obviously not selfish. People link to each other all the time precisely because they think others are worth paying attention to, and that seems to me to be the right standard.

To abide by a principle of informational efficiency is to not want eyeballs for eyeballs sake, but to want eyeballs to be directed to what is the most important, whether it be on one's own site or elsewhere. If you treat me according to that standard, that is the only reciprocity I want. I don't want people looking at my posts when they can find better understanding elsewhere. If everybody uses their best judgment about how best to direct each other's attention, that is ideal.

Chris also thinks that the idea of not linking to crap is censorious, and I am inclined to agree. Chris writes:
Bluntly speaking, you're acting as a censor. Who are you to make the decision that something is crap? If you only provide a link when you and the person you're linking to agree, you're not providing efficiency - you're creating an echo chamber. What I'm doing is discouraging censorious behavior and encouraging debate by withdrawing linkage that does not reciprocate, regardless if the person is agreeing with my position or not.

Spot on. But I don't think that is inconsistent with the standard of informational efficiency that I am advocating. Chris recognizes the value of linking to contrary opinons and so do I. So why did I raise the example of not linking to crap? I hadn't thought it through. Crap can be good (to READ, even if the million flies as Kos and Talking Points ARE all wrong). So set aside the business about not linking to crap. I take it back, and it was a side-issue to begin with. The main reason I am in favor of assymetry in placing trackbacks is because placing trackbacks is work, and there is assymetry in the value. As a result, full accounting of value will generally lead to assymetry in linking.

As for Chris' concern about spam, if the TrackArosses were administered by HaloScan just like TrackBacks, there would be no additional spam risk, except that what spam does get through on TrackBacks could be interated on TrackAcrosses. (Chris hadn't seen my post above about entering TrackAcrosses in a second HaloScan window, so he was not being obtuse to think that TrackArosses could add a whole other spam headache.)

The other concern of Chris's that I think is important is his bandwith concern, not gigabyte bandwidth but eyeball bandwidth. Obviously Chris thinks that what he writes is worth paying attention to, so he wants to give pride of place those links that comment on his contributions. The correct answer I think is to have TrackBacks and TrackArosses under separate links and let readers look at whichever they want.

My final thought for the night: before too long somebody will automate the option of inserting "x,y and z are also talking about this issue" in their posts when TrackBacks are left. At that point there will no longer be a labor issue. The only question will be how much help it is to readers to be able to distinguish between linked posts that actually comment on what they just read and posts that simply self-cross-reference it. Will that distinction be valuable enough to mark with a separate button? Maybe.

Wow, it's not much better in daylight. Let me try more coffee.... :-)
Post a Comment

<< Home

This page is powered by Blogger. Isn't yours?