off.Root

(Sidenote: I originally was going to title this entry “Nofollow: destroyer of the Intraweb…?” but decided the new title was more fitting.)

I’ve been (somewhat) on the fence about the whole rel=”nofollow” thing, as not-so-recently announced by Google and a couple friends. Here’s a draft spec, for the uninitiated. But a lot of people who are gung-ho about nofollow being the single worst thing to ever happen to the Internet need to reevaluate some of their views. Perhaps their conclusion (nofollow is more harm than good) isn’t necessarily inaccurate, but their assumptions most certainly are.

It’d seem to me that in order to declare using rel=”nofollow” on links that can’t be trusted officially “bad,” it would have to be proven that nofollow would actually worsen the current situation with comment and trackback spam. And few people have seemed to be able to make that argument. Most who say it’s a bad thing say so because it breaks the Internet, destroys the “web” of links within the blogosphere, or simply wouldn’t stop the efforts of spammers in the first place. And of course, there are those who swear that this is just another ploy by Google to get it closer to world-domination. I have my doubts about at least that last point, but the problem with these arguments is either that they’re untrue or don’t make nofollow “bad.” Ineffective, perhaps. Not “bad.”

Why is there the (theoretical) need for nofollow? Because the foundations of the Internet are already being attacked by spammers. Search engines have relied on links pointing to a given page to determine its popularity, and now with spammers abusing these methods it’s time to find a workaround. Was it a flaw in the conception of search engines? I don’t know. The Internet back then was smaller and less commercial. If a search engine saw a link on a web page it was a safe assumption that that was put there by the author and owner of the page, and as such it was an endorsement of the link as far as PageRank goes. This is no longer true with the growth of interactive sections of the Internet, like blogs, message boards, guestbooks, and the like. Which begs the question: to what else will nofollow be applied? Will we see support for it in the next version of phpBB or vBulletin (or that “other software”)? Probably not, given that most people have adapted the mindset that links have the primary purpose of increasing search ranking, and the secondary purpose of linking to another page.

But back to the blogs. First, to dispel some of the myths. Nofollow will not destroy the Internet. All nofollow does is send a shoutout to search engines that a given link appearing on a page wasn’t, contrary to common perception among search engines, put there by the author, and as such he doesn’t want you to consider the link as an endorsement. Because Google also penalizes sites linking to so-called “bad neighborhoods,” it’s really (sadly) in the best interest of the webmaster to implement it as to avoid being penalized for any spam that does get through.

That said, nofollow - if implemented correctly - will not affect the natural web of links on the Internet. The end user won’t know without checking the source whether or not a link has nofollow on it, and will be able to follow the link however they please, spam or not. Links still show up, and for the end user they still work. And given that implementations of nofollow will apply only to third-party generated content (comments/[track|ping]backs), nothing is forcing you to apply nofollow to your blogroll, or stopping you from giving credit to those links you think actually deserve credit.

By the same token, having nofollow on any given blog won’t mean that a rash of links to casinos and cheap mortgages won’t show up, but nofollow is simply one tool among a long list of tools. Install your other spam plugins, those’ll help too. If nofollow doesn’t curb the number of spammers, something even I don’t consider especially likely, then it can at worst (based on the above) be considered “somewhat ineffective,” not “bad.” Simply consider why spam exists (money!): junk e-mail spammers want to get their message to you. If you implement spam filters, you’ll never see it, and in the long run one hopes spammers won’t bother anymore. But who stands out and shouts that filters are no good because not everyone’s grandma has one? Who complains that they should be disabled because spammers will just find another way to get their message across? The messages are still sent, the bandwidth is still used, and the build-up will just increase.

But nofollow is a bad thing, because spammers will just figure out how to get around it. Spammers will keep spamming - but that’s always true, it’s not as if they’re sitting on their asses all the time waiting for a reason to blast a few million more sites. Why don’t you disable your captchas, then? Machines are getting pretty good at deciphering them, and they might get better unless people stop using them.

Yeah, see the logic? Captchas at least block off a portion of your visitors (the blind, those who don’t use image-enabled browsers).

The thing about comment spam is that most of it is not to “reach an audience,” because a lot of the time the spam is so obvious to a human that they wouldn’t waste their time with it. Boosting their search results is a real goal, and if it proves ineffective in the long run, it’ll stop being a viable option. Granted, it will probably just result in increased long product pitches on blogs, but these will reach smaller audiences than search engines and (hopefully) make it less attractive. In any event, it’s certainly not an instant-off tactic, and it’s not something that’ll curb comment spam on its own. (This is, apparently, the one exception to the “links are for search engines not for people” concept I mentioned above.)

And to be frank, as general spam-filtering tactics increase, comment spam will likely decrease because it’ll become too hard to get too few comments through. Just take a quick perusal of the list of Ping-o-Matic’s recent pingers and you’ll see over a short period of time there are plenty of spamsites out there masquerading as blogs. They filter into Technorati and Feedster and whatnot, and then their message is placed and they have their exposure, seeping into search engines all the while. Spam is going to evolve no matter what you do about it, and they’ll find more effective tactics to get traffic.

But a lot of the problem is implementation, implementation, implementation. A lot of the fault here that people see doesn’t fall on Google & Co., it falls on how they see nofollow will be used in practice. And there are some flaws to how I see nofollow will likely be implemented, and how certain blogging services will handle it.

Nofollow shouldn’t be a blanket-it thing. WordPress has committed to including it, which is a good thing - just not how they do it. In an interview, Matt (lead WP developer who, last I checked, has nofollow enabled on his comment links) says WP will include nofollow, but users should be able to remove it thanks to WP’s modularity. (Really, he doesn’t say that, he just says users can change the default behavior if they can “intelligently make a decision” about it.)

But, jab at nofollow-supporters aside, why does it come down to that? WordPress in particular is awful at including features that the end-user might not want, but providing no means of actually disabling them. The whole curly-quotes thing, while I don’t mind it, is certainly irksome to certain users. Occasionally it seems WP also morphs my attempts at including actual comments in my posts, which is frustrating to the umpteenth degree, and frankly not something I should see from software that says it’s what to use when you want to avoid fighting with your weblog software. The “less is more” mentality is okay (it works for Firefox, and it almost worked for phpBB, at least before they caved in to making their next major release a feature-packed 3.0), except it’s based on the idea of adding plugins/extensions both easily (check, with one-click installs), and to change expected behavior. Curly quotes and nofollow don’t really fall under “expected” behavior under a blog. That doesn’t mean they can’t be enabled by default, or shouldn’t be, but if a user has to go elsewhere to remove unexpected behavior (like, many would argue, adding seemingly random but semantically correct <li> elements) then there’s a problem. Those features are the ones that absolutely must be made options, and I really wish people would drop the “customization through options is bloat” mentality, and stop trying to convince users that they don’t want to be able to change unexpected behavior out of the box.

I would like to implement nofollow on my site (and since it’s default in WP 1.5, I believe I have), but there are limitations to how WordPress has implemented it that makes it more frustrating. This solution isn’t perfect either, but it makes much more sense than WP’s default can’t-touch-this setting. There is no way, as an option in the Admin panel, to modify the nofollow behavior, and I don’t understand why. Sure, enable it by default, but why tell an author that they have to have it enabled? The temporary nofollow plugin, by the way, makes it so that nofollow only applies to comments that are less than 10 days old - the idea being that you’ll have deleted bad comments by then, so anything that old is safe. Makes sense to me, and it lets me credit comment links that I deem valid.

What would I like to see? A lot of it really limits spam in general. Establish trusted users, make it easier to let users register as commenters if they so please. If I post a link in my own comments section I don’t know why it should get nofollow’d. Let me okay links like I would comments, so that even if it’s posted I can still remove the nofollow at least on a comment-by-comment basis. In an ideal world these links won’t sit around for long enough for nofollow to matter in the first place. It’s very much a temporary solution, in more than one sense of the word.

But please, please, please don’t try to tell me what’s best for me. I remember Gaim had a period where it seemed the developers were trying to do everything they could to take the program many steps backwards, and one of the things they did that annoyed me the most was removing a bunch of preferences and replacing them with default values - a bunch of which, I noticed, were not values I wanted or had set. It got annoying when chat rooms inexplicably started showing up in the same window as IM logs, and I had no way of changing it (and it still bothers me when it decides that a user who sends me a message to which I respond is a “buddy” of mine and silently adds him to my buddy list). By all means, come up with a set of default options to help the user without enough knowledge, or for the user who doesn’t want to bother. But don’t tell us what to do and tell us you’re protecting us from ourselves.

So. Is nofollow a good thing? Who knows. I don’t see it as an effective tool against decreasing comment spam, at least not in the near to distant future. But I don’t see it as especially harmful - if developers can figure out how to get it right. Google’s done its part getting this out the door, but everyone else has to step up and make it work. Just slapping rel=”nofollow” on any comment is nice and easy to do, but it’s the wrong way to do it.