Saturday 26 February 2011

What's Really Wrong with SEO?

I have little sympathy for companies who've been doing SEO and are now finding it not working and I find it most surprising that people are objecting to Google's recent changes. So here I try to spell it out in the most elementary terms what is wrong in principle with SEO as a practice.

Writing content to improve your search engine rank is like teaching to the test. It's letting that which is supposed to be measured, the quality of your webpage, be shaped by the metric, its position in search results pages, rather than the qualities the metric is designed to measure. Good metrics should be indicators of success, but that can be undermined when there is a focus solely on the metric at the expense of that which the metric is supposed to measure. If your scores are improving as you pursue the real objective, e.g., good authoritative content or clear understanding of the subject material, that's very likely as it should be. And if you're successfully chasing the objective and the metrics don't reflect it, possibly so much the worse for the metric.

But one should never make the metric itself the primary goal. The reasons should be obvious. One may very well end up over-fitting to the metric and gaining "improvements" because one has keyed in on secondary or indirect or ephemeral, but easily quantified, features of the quality under consideration. As the metric becomes better understood by those being measured, those indirect features are then purposefully exaggerated and, as a result, no longer act as reliable indicators of what they once indicated. Consider link text, the feature that Google used to use so heavily. At one time it was a great indicator of popularity. But as soon as people realized that Google perceived it as an indicator of popularity, they started using it to create the impression of popularity and it ended up becoming a far less reliable indicator.  Of course, Google had to change the extent to which they relied on it to measure popularity. This is inevitable when everyone attempts to game Google.  And we shouldn't be surprised or complain when Google changes their measurement tools in response to a change in the characteristics that they're trying to measure.

As an example of what I'm talking about, I read this objection to Google's attempt to penalize content farms. The author lists some methods of SEO:

* Research keywords
* Select keywords that have existing traffic
* Write pages based on those keywords
* Publish pages
* Get those pages ranked against those keywords


S/he goes on to ask, "How is this different to what a Content Farm does? So, if Content Farm pages are undesirable, so too is SEO content?"  (apparently "content farm" is worthy of capitalization!) Can you imagine the great writers of history using anything like these techniques for writing high quality material?  These aren't strategies indicative of someone trying to write useful, relevant, clear content; they're strategies for people trying to be manipulative, advertisers or propagandists. Insofar as people prefer not to get advertisements or propaganda in their search results, Google is right to penalize these methods. Rather than whining, people should go back to thinking about how to write web pages people want to read and let Google worry about ranking them.

Friday 25 February 2011

Remember when Reich wasn't Radical?

Robert Reich has put up a couple of posts recently on the wealth inequalities in the US and what they mean for the US budget.  (An aside, the Mother Jones graphs are telling but nothing, to my mind, is a more bracing summary of the situation than the fact that the 400 richest Americans have nearly as much combined wealth  as the cumulative wealth of the bottom 50%.) 

The first post  discusses the extent to which changes in income distribution have affected Social Security and how it's just untrue that Social Security contributes to the deficit. In 1983 a Greenspan commission tried to fix Social Security by implementing a plan whereby the ceiling, the point after which wages don't require a SS contribute, would continue to grow gradually, "the ceiling was set so the Social Security payroll tax would hit 90 percent of all wages covered by Social Security."  However, the wage inequalities have been such that the current ceiling of 106,800 covers only 84% of all wages.  To get it back up to the 90% level we'd have to increase it $180 000!  That would be an easy fix for SS and seems a small price to pay for those benefiting from the wealth inequality, but as Reich notes, this is apparently unacceptable in the present climate.

The second post is a bit less wonkish, simply noting the BS that the Republicans are serving up with respect to our economic travails and the Democrats' collective lack of willingness to call them on it.  The "Republican message is bloated government is responsible for the lousy economy that most people continue to experience. Cut the bloat and jobs and wages will return." Of course, the GOP assumptions are highly questionable, as Goldman Sachs notes. But the problem, Reich notes, is that the GOP is now allowed to get away with this stuff. We don't hear the Democrats or Obama talking about the obscene inequities in income distribution or dispelling the myth that government is responsible for the economic difficulties we're facing.  Instead, Reich observes, the Democrats simply counter with "We agree but you’re going too far. Government employees should give up some more wages and benefits but don’t take away their bargaining rights. ... Non-defense discretionary spending should be cut but don’t cut so much." I admit that I've been pulled into this game too.  It's yet another example of the GOP strategy to pull everything hard to the right and then make even conservative positions look left wing. They've played it marvellously in Wisconsin.  Roll back union benefits?  To hell with that, we'll take away their collective bargaining power too. Suddenly the left is fighting just to retain collective bargaining and we've sold the unions down the river on their compensation packages, unquestioningly buying into the myth that they're egregiously overcompensated, even while we blithely accept ludicrous executive compensation packages as a natural and even desirable part of capitalism.

What's interesting and alarming is the extent to which Reich's positions are now radical, far outside the extant political mainstream. Recall that it was only two administrations ago that Reich held a position in the Cabinet. We've come a long way, baby.

Wednesday 9 February 2011

Tea Party comes to Washington to fight for liberty?

There's a story in this morning's WaPo about the Tea Party allying with Democrats to help defeat an extension of the Patriot Act.  One reads that it was "a vote that served as the first small uprising of the party's tea-party bloc." I was pleasantly surprised initially. My impression of the Tea Party has been that they're more of an ultra-conservative group of xenophobes than a group seriously interested or concerned about liberty. Tea Party activists, for example, came out in strong support of the Arizona immigration crackdown, real libertarians aren't inclined to have much truck with opposing immigration.  It would be nice if the Tea Partiers were actually taking civil liberties seriously, if they were more Ron Paul-like and less Sarah Palin/Michelle Bachmanm-esque, I'd be less inclined to think the Tea Party was bad for America. 

In any event, my hopes were quickly dashed when I dug into the numbers a bit.  This alleged pro-liberty Tea Party uprising was exaggerated significantly by the Post.  Of 111 congressmen that the Tea Party had endorsed, 96 (86%) voted in favour of the Patriot Act extension, only 12 (11%) voted against and three didn't vote.  Yeah, sure, that's some uprising.

Wednesday 2 February 2011

More on sodium thiopental

Ah, some more curious moral reasoning regarding Hospira's refusal to continue manufacturing sodium thiopental, (see earlier comments).   Charles Lane objects to the discontinuation of its production because (i) there are legitimate anaesthetic uses of the drug and (ii) "I suppose European restrictions on thiopental might be justifiable if they save a lot of lives on Death Row. They probably won't."

Consider the second point first. Lane seems to be contending that it's justifiable to refuse to contribute to an objectionable moral act only if you have good reason to think that your refusal to participate will prevent the act from occurring.  It's okay (required?) to contribute to morally objcectionable acts if it's reasonable to think that your contribution to the act is relatively easy to replicate or replace.  Suppose I'm producing some drug that turns out to be highly effective for date rape and is, in fact, often used for that purpose. Suppose further that for that reason I stop producing it.  Well, according to Lane I'm not justified in doing so because, after all, if people don't have my drug they can always use booze or roofies or something to accomplish the same goal. 

Let's also consider the objection that there are legitimate uses of sodium thiopental and let's assume that that's true.  Where does the moral compulsion here lie -- on the producer of the drug or on the ones intentionally using the drug for unintended purposes?   Suppose, again, that I'm producing a drug that can be used for date rape but also has legitimate uses.  Suppose you are in control of distributing the drug and I ask you not to sell it to people who tell you that they're planning to use it for date rape.  Suppose, you refuse this request.   And suppose that because of your unwillingness to provide this assurance, I refuse to continue to produce it.  Who is to blame here for the fact that some are unable to use the drug for its intended end, those who who refused to produce it because of the reasonable belief that it will be misused or those who refused to use it for its intended purpose?