trenchant.org

by adam mathes
archive · subscribe

Comments Gone Wild

Matt recently wrote about a phenomenon that might be called “Comments Gone Wild.”

These are weblog posts that started out as a simple post by a weblogger about a television show. Then readers start posting comments that take a beautifully comic and yet saddening turn as confused visitors think they are commenting directly to the show.

The first post from blog.noetech.com concerns the show “Overhaulin”, the other from laze.net concerns Maury Povich.

From Matt’s commentary:

"It seems obvious to all us that have seen weblogs before, the original post clearly is someone commenting on a TV show. But my guess is that regular folks see Google as an internet appliance, and when you put in "overhaulin" you will get the right site as the first result and if that site asks for comments, it must be the show, right?"

More from Matt:

"Aside from my Google hypothesis, my other one of course is that people never read anything and Google results leading to blogs just makes that more obvious."

While I don’t dispute Matt’s claims about user stupidity and lack of reading, I think a more thorough analysis may be more helpful and interesting in evaluating how this happened. This is not simply about whether people have seen weblogs before or not.

I see this as an information design failure on multiple levels.

Google’s Failure

These casual visitors didn’t just happen upon these sites from random surfing: they were directed there by Google queries, presumably. Or other search engines, I don’t mean to pick on Google here. (Well, maybe I do.) These one-word queries returned relevant initial results in some sense, but clearly the first result in these specific cases was not pertinent to the majority of users queries.

That is, most people who search for “Overhaulin” aren’t interested in a weblog writer’s thoughts on the show, they are actually looking for official information about the show. Normally it is difficult to determine what a user is really looking for (their ‘query’) from just their search statements (the individual words they used to search) but in this case we know exactly what they were looking for since they spell it in excruciating detail in the comments they left.

But there’s not much you can do about Google sending you visitors for search terms that don’t accurately reflect your content. There is no such field as search engine de-optimization as far as I know; you either let the spiders crawl you and show up in search results or don’t. Yes, robots.txt and htaccess and other tools can help determine which parts of your site are crawled and who can crawl them, but in the end, it’s almost impossible to control completely which, if any, terms your page will be returned as a result for. (Also, it’s not just your content that matters, links are key.)

I’m not really of the belief that weblogs and personal sites “clutter” up the web or search results. Often, personal commentary by a domain expert may be better than other more “mainstream” results for a query. But regardless, this is in no way the fault of web writers, this is a search engine problem. And as far as I’m concerned, they can solve their own problems.

Failure of Information Architecture

What is this page? Where am I? Who wrote this site? These things need to be apparent for a visitor who shows up to any individual page on the site. They need context. While I’m not saying the pages in question don’t do a good job of that, it’s an important aspect to keep in mind. (It’s one of the things I’ve been thinking about for my pending real-soon-now redesign of trenchant.org.)

Examine the page from this perspective:

What is this page about?

As comments overwhelm the original content of weblog posts, the breakdown of visual space on the page lends credence to the sense that this is a place where people write “to maury” or “to overhaulin.” Somebody who arrived at these pages from a search engine may simply skim and pass over the original content of that post. After just one screen of scrolling, there are no contextual clues to explain what the page is. There are only comments from people that begin with Dear Overhaulin.

If you want your page to be “accessible” to search engine visitors you should think about this in the design of your site at the page level and the site level. Do you want comments to be able to overwhelm your content? Maybe you want a discussion forum elsewhere on the site (that nobody will use.)

Failure Of Community

Whether or not one views their personal web site as a community, offering the ability for users to leave comments on your work inadvertently creates the possibility for community. In the case of some software, a community with almost no barrier to entry other than stumbling upon a single page of the site. What kind of barrier to entry to have for a community is something worth thinking about. Derek Powazek devotes an entire chapter in Design for Community to barriers to entry, concluding with:

"Sometimes, the best thing you can do for your community is to make it just slightly hard to reach. And while you may not want to take it all down dramatically or hold a village burning, you should pay close attention to the barrier to entry on your site. Nudging it up or down with care may be the most important piece of design you create."

(See, Derek, I read your book!)

Having a site with open comments, combined with weird search results, is a little like having acquaintances you don’t really know pass out flyers that say your house is the best place to go for “lesbian sailor moon scouts.” (A not uncommon search term for my site.) If somebody shows up for that, do you really want them to find not just an open door, but a can of spraypaint and big wall filled with “Sailor Neptune 4eva!!” on it?

It’s funny how everything always comes back to Sailor Moon. Err, I mean, badly designed personal publishing tools.

Failure of Tools

I don’t think it’s merely chance that both of the examples cited here are weblogs that are published with

Movable Type.

Current popular weblog tools do a questionable job at making sites into communities with comments on posts. The notion that “every entry is a start to a conversation” is a little odd to me, but it seems to have gained traction. I have long questioned in particular, the default Movable Type configuration that (to the best of my knowledge) has allowed unmoderated, unregistered, anonymous comments on any entry at any time. This just seems irresponsible. I am not sure that most people who started a Movable Type weblog consciously accepted and wanted this arrangement. My understand is that this is changing with Movable Type 3.0, and that is good, but with a large installed base of users, and the licensing changes in 3.0 I still think it is worthwhile to examine the consequences of the design decisions of previous version of the software.

In my view, any software that allows content to be added to your site without at least your knowledge, and preferably your approval, is dangerous. It allows spammers to hijack your site without knowing by flooding your site with advertisements. It allows trolls to destroy discussion. It allows material that you may not think is appropriate or relevant to be hosted on your site.

While tools to edit and moderate comments are helpful, and individually approving each comment may not be feasible for large sites, but if I were to design a small content management system for weblogs, I would make the default more rather than less restrictive.

Finally, there is, also, the obvious responsibility of the individual site owners to monitor what happens to their site. But writing about personal responsibility is boring, and I don’t want to put the few readers I have left asleep.

My point is, just deriding casual internet users as simply looking at Google as an “internet appliance” is not in my opinion the best way to examine this situation. A better analysis might be, due to some past success with Google, users have high expectations for relevant search results, and when the sites aren’t pertinent to their query, they may not immediately recognize that if the resulting pages fail to provide enough context. If those pages made poor design and community decisions, their confusion may end up directly on those pages.

This is probably a situation that will become more common as search results get worse and the web grows with users who can’t be bothered to read the headers of pages.

· · ·

If you enjoyed this post, please join my mailing list