It’s kind of a thing

August 30, 2002

Google: Undemocratic?

It seems that another person has come forward to question Google’s impartiality. According to an article at Salon, one Daniel Brandt believes that Google’s PageRank algorithm is biased and undemocratic.

A quick refresher: PageRank is one of several methods Google uses to evaluate pages in response to a query. The exact details are secret, but it involves calculating how important a page is based on the importance of the pages which link to it. As I recall from an old paper written about their early implementations, it involves lots of matrix manipulations that get applied repeatedly until things stabilize. Google claims that this method is unbiased because it is completely automated; pages with a high PageRank (apparently named for Larry Page and not the fact that it ranks pages) are those which the web as a whole have deemed linkworthy. Furthermore, it’s not the only factor considered. ZedneWeb, for example, has a pretty low PageRank (so far as I’m aware), but is listed pretty high in searches for “David Menendez” because my name is mentioned in the page title. While PageRank can occasionally place an important but irrelevant site high in the results, it’s mostly pretty solid.

Mr Brandt would no doubt disagree with my assessment. I won’t claim to know what he’s thinking and I haven’t visited his site google-watch.com, but his main complaint seems to stem from a misunderstanding of Google’s purpose. Part of the reason Google became so popular was because people could ask it for IBM and the first entry would be IBM’s site (plus, all the other results would be high-profile, non-porn sites that mention IBM—I remember some of Google’s promotional material where they made a big deal of the fact that every result they returned actually contained the search terms; doesn’t give one a good impression of their competitors, does it). Mr Brandt would probably prefer some site that promotes awareness of IBM’s awesome power and nefarious schemes.

(I’m not slamming IBM, by the way. I just picked a big company at random.)

On the other hand, he does raise an issue that I had not been aware of: Google’s 36-year cookie. Apparently, Google uses a super-long-term cookie to associate a user with their search preferences. That seems pretty innocuous, but consider that this means Google can also keep a list of the things you search for. Again, that’s not a bad thing in itself, but it does give one pause.

Fortunately, you can use Google without cookies and can delete the cookie without causing problems. Also, because Google doesn’t collect information like your name or e-mail address, they don’t have any real way to link your searching history to you personally. And the vast number of preferences in their database and the huge number of queries means it’s unlikely that they’re paying much attention to you, personally. But still….

I’m not worried. It seems unlikely that Google has sinister intentions at this point. Perhaps new evidence will emerge, but until then there’s no point in avoiding the most effective search engine on the web #

Hot off the wires

I’ve played around with NetNewsWire Lite a bit since I last mentioned it, and I can confidently say that it’s very well made. It’s fast and reliable and has a pretty straightforward interface. The only problem, for me, is that it doesn’t fit into my browsing style.

Essentially, NetNewsWire is a web browser companion optimized for reading the headlines of multiple web sites (assuming they provide an RSS feed). Looking up the headlines in NetNewsWire has two advantages over looking them up in the web browser. It’s faster, because it involves downloading smaller files, and it remembers which headlines you’ve already seen. Thus, someone who has subscribed to ZedneWeb’s headline feed can see at a glance when I’ve added new things, because NetNewsWire will note that there are X unread posts.

Thing is, I prefer reading weblogs directly, as they tend to involve lots of short articles on the same page. It makes more sense for larger, magazine-like sites and those Slashdot-style news forums, but I don’t read as many of them as I used to.

On the other hand, I’ve been exposed to some interesting stuff that I wouldn’t have otherwise seen if it weren’t for NetNewsWire’s default subscriptions.

For example, an article at Plastic notes the possibility that MIT used images from the comic book Radix in their bid for a military contract to design equipment for the high-tech soldier of the future.

Or this Kuro5hin article about political clichés in sci-fi/fantasy settings, particularly the way people accept monarchies, chosen ones, and us-versus-them morality in these stories without thinking about them too much. I actually think the article’s arguments aren’t particularly strong, but it does touch on a point relating to my own work.

Matt Groening, talking about the ill-fated classic Futurama, noted that one thing he was trying to avoid was the new age military feel seen on shows like, say, Star Trek. (Which, if you look closely enough, seems to be advocating socialist military dictatorship as the ideal government.) That comment struck a chord with me, because it made me recognize an overly-military feel in Starcruiser Anonymous, where virtually everyone has a rank. Although I haven’t had much chance to demonstrate it, I have chosen to downplay rank in the future—actually, I’ve chosen to pretend those references were never made—and give the crew more of a private sector feel.

But getting back to NetNewsWire—the theoretical subject of this post—it does illustrate another reason to break up the web browser. Right now, your web browser keeps a list of all the pages you’ve visited in the recent past (for some value of “recent”). NetNewsWire also keeps track of pages you’ve visited, in the sense that it remembers which items you chose to view in the browser. Unfortunately, it has no way of knowing what pages you’ve visited in the browser on your own. If both NetNewsWire and your browser(s) used a common history, that, at least would not be an issue. Not a compelling reason by itself, just something nice. #

Avoiding the flame war

There was a recent minor flare up in a part of the weblog community. Rather than describe it myself, I’ll refer you to Jonathon Delacour’s summary which pretty much explains everything and links to the original posts so you can see what happened yourself.

What’s notable is how civil the discussion was. Had similar accusations been made on a web board or on Usenet, it’s likely that both sides would have started namecalling and created a big ugly stink until someone apologized or everyone involved got bored. That’s a huge generalization—plenty of news groups and message boards manage to avoid generating into flame wars—and this was probably a special case, as the people involved are fairly high-profile and generally civil, but it interests me.

One theory that’s been proposed to explain the tendency of on-line discussions to devolve into pointless argument has to do with the anonymity of on-line discussion. This sort of argument is more likely to occur among people who don’t know each other very well (or actively dislike each other). Another theory simply notes that many of the cues present when people speak in person, or even over the phone, are absent in text. It’s easier to misinterpret what someone meant to say, because you can’t go by facial expressions, gestures, or tone of voice. Irony, in particular, is a minefield on-line.

But none of that explains why weblogs don’t have the flame wars that you see in other forums. In fact, there’s no hard evidence that they don’t; I’m basing this all on my impressions. But since I’m on a roll, I’ll continue.

The primary difference between inter-weblog discussion and other forums is that weblog posts are naturally grouped by author, while forums are grouped by topic. While my thread description language is able to treat them identically, I think the different organization changes the dynamic of conversations, because they’re harder work. On a message board or news group, replying to someone is simple. In the blogosphere, the barrier to entry is higher, because it requires you to have a weblog. Then, replying to someone is a matter of referring to their post in your own—but that doesn’t guarantee that the person you’re replying to will even see your post. Carrying on a back-and-forth discussion requires both sides to pay attention to each other. It’s harder for third parties to enter and turn the argument into a real flame war, because they have to be noticed as well.

Of course, many weblogs have commenting systems, and I’ve seen plenty of rudeness in those. In fact, I’m sure there have been back-and-forth flames among weblogs. But something about the structure of the blogosphere makes it seem somewhat flame-retardant to me. I suppose time will tell. (via Doc Searls) #


From C-net: “Games push for more than pretty faces”. As the increases in processing power and software complexity allow more and more realistic-looking video games, designers are starting to worry about other aspects of realism, like non-player characters who aren’t morons. While I remain impressed with what’s possible visually these days, I also remember that visuals and sophistication aren’t everything. After all, people still talk about Marathon even though its engine is laughably primitive today, because it had such a great story and fun netplay. (Which, come to think of it, have nothing to do with each other. Weird.) #