Steve Reads

Norman Mailer: writes long, rambly, pretentious pseudo-journalism. Obviously thinks he’s the best thing to happen to literature since the invention of the semicolon. May be remembered in 50 years for The Executioner’s Song, but probably not; it will probably go into the same category as In Cold Blood — historically interesting but ultimately not all that timeless. Will certainly not be remembered for Harlot’s Ghost.

Michiko Kakutani: forceful, passionate book critic for the New York Times. Will be remembered as long as anyone remembers criticism (which isn’t, honestly, probably all that long, unless you’re Pauline Kael).

Now they’re in a fight, which Michiko is winning without (indeed, because she’s not) saying anything.

(Via The Bookslut)

I just noticed another bit about open-source software that is uncommon — if not nonexistent — among commercial software companies: you quite often talk directly with the inventor of a particular open-source package. Rael Dornfest, the creator of the blog software on which this site runs, is quite often on the Blosxom mailing list. The inventor of vim, Bram Moolenaar, is on the vim-dev mailing list every day. This shows up again and again throughout the open-source world.

Closed-source software companies, as a general rule, don’t let their developers speak with the public (though I’ve heard of a few exceptions). Software companies leave themselves open to too much liability that way. If one of your developers says something about how a particular feature is very buggy and prone to leave your system unstable, and it so happens that the software loses one of your clients a great deal of data some months later, my sense is that the developer’s comments provide evidence in any suit against you. That’s one of the big reasons why closed-source companies tend to limit their public communications to designated people: PR, tech support, etc.

It’s fun to talk with actual developers. You feel like your suggestions are actually going to go somewhere — which, in fact, they are, most of the time.

Spain now has gay marriage. Canada approved it the other day. Presumably more nations will follow.

If U.S. states pass gay marriage one by one, I think I’d be fine with that. In a country this large, we can’t expect that every state — some of which have nearly as many people as Spain — will come to support gay marriage at the same time. Strategically, it also seems smarter to do it piecemeal; we may get what we want faster that way, with less acrimony than if we insisted on pushing gay marriage through Congress.

That said, I do believe it’s an issue of fundamental equalities. But I also believe that the U.S. is probably not ready to allow gay marriage at a national level, and fundamental equalities may — for now — have to take a backseat to pragmatism.

The case of “Dog Shit Girl” in Korea is really interesting to me. I’m very tired right now, so I don’t really have much to write. But

1. Richard Posner has noted that privacy — overbroadly applied — is often used to cover up acts that it would be valuable for society to know. It would certainly be valuable to know that the person you’re dealing with is a habitual sociopath.

2. As others have pointed out, there are some good social benefits from publishing this woman’s misbehavior. Probably a lot of people will now think twice before behaving awfully in public.

3. This is supposed to be evidence of how the Internet has changed everything, but that’s not clear to me. I’m sure Richard Jewell — the suspected “Atlanta Park Bomber,” whose name I remember today years after he was exonerated — would argue quite passionately that the major media are as bad as, if not worse than, the Net.

4. Not all vigilantism is created equal. Some “vigilantism” is just social norms with teeth. Which is to say that the state isn’t the only legitimate source of enforcement. I think we should be more concerned about vigilantism that leads to, say, death. That’s not the final standard that we should land on, and of course the line is hard to draw (what happens if my posse’s Net vigilantism leads to someone’s real-life vigilantism, and someone dies as a result?). But certainly not all types of non-governmental action deserve the same presumption of scorn.

5. Probably there’s more to say, but that will do for now.

It’s been a while since I’ve sat down to use IM, but I happen to be on today for business purposes, and I’m using gaim for the first time in probably a couple years. It’s awfully good. Wow. It was quite a bit better than just “functional” when I last used it, but now it’s actually polished. It’s a very clean, usable, intuitive app. I’m quite impressed.

I like Debian as much as the next guy, but I think tattooing the Debian logo on yourself takes it just a bit far.

Google keeps doing all the right stuff. Now they’ve released a Google Maps API — meaning that if you want to fold Google Maps into your programs, you don’t need to spend time hacking it to, say, figure out the format of Google Maps URLs. Now you can just use their API. So. Fucking. Cool.

Quick thought: if Google included a checkbox beside every link that it returned to you, and checking that box flagged the link for everyone else, they’d basically have as much information as del.icio.us does.

Del.icio.us seems kind of superfluous in the presence of Google. There are a number of services trying to derive information of one sort or another from links: Technorati and del.icio.us are probably the most well-known, apart from Google. But Google makes all of them redundant, or at least could very easily. Google knows

1. Which pages are most highly linked.
2. Which pages people are clicking on.
3. Which pages are most popular among those people who like various other pages (i.e., Google knows how to cluster web pages by popularity, like Amazon does with books)
4. Which links people are emailing one another (via Gmail).
5. Which links people are blogging about (via Blogger).
6. Which links are most popular in specific subcategories like news (via Google News).

Google is the world’s most comprehensive information sink for information about which pages are popular. As long as Google retains its hegemony here, it seems silly to even think about competing with them on link information.

P.S.: The fact that a user doesn’t click on a link is highly informative. If you get 10 pages full of links that aren’t relevant to your query, and you indicate to Google — either by clicking a checkbox labeled ‘irrelevant’ or by simply not clicking — that those links are no good, I’m sure Google could do a lot to improve your queries. This would be particularly good, I wager, to combat services that try to defeat PageRank by creating a lot of bogus sites and linking amongst them. I’m sure Google has done a lot of research to weed out these services by establishing which ones are in self-linking cliques, but it would still be useful to have users validate which sites are junk.

Via Del.icio.us, I see that Bertrand Russell’s essay “Why I Am Not A Christian” has been linked a lot recently.

The rest of the book containing “Why I Am Not A Christian” is rather fatuous, but the essay itself is actually a very straight-ahead look at the logical and evidence-based reasons for belief in Christianity. Basically: if you accept some scientific results about the structure of the universe, and you accept basic laws of logic, the arguments normally evinced for god’s existence don’t make any sense. This sentence in particular does good things for me: “Do you think that, if you were granted omnipotence and omniscience and millions of years in which to perfect your world, you could produce nothing better than the Ku Klux Klan or the Fascists?”

Russell is probably right that Christianity cannot be justified by appeals to argument and evidence. So what’s left? Is there any reason why a scientific person, looking at this sometimes very ugly world, should believe in God?

To many of us with a scientific worldview, it seems as though Christianity has been proven wrong every time it deigns to make a factual assertion. Humanity is not at the center of the universe, as far as anyone can tell. God appears not to have designed biological organisms; if there is a god, as far as anyone can tell, he set certain processes in motion and let them proceed in their own imperfect way. God seems not to be interventionist; the Universe, again, seems to have been designed with a great deal of predictability, and any neutral standard of evidence seems to weigh against miracles.

So to many of us, the place for Christianity appears to have been pushed back. Its explanation of the Universe has been forced back to arguing that God set certain processes in motion — hence that evolution and the Big Bang are not inconsistent with God’s existence, and that it is merely humanity’s role to understand His mind in creating these processes.

Christianity also seems to fill a role as a moral authority, in a way that science has perhaps not done — though it seems to me that deciding what your goals are (say, easing the plight of the world’s poor) and then trying to find the best way to achieve those goals (e.g., by appropriate fiscal policy) would do as much to fix humanity’s problems as would adopting a religious outlook. This leaves open the question of how one sets those goals, but I see no reason why we need religion for that purpose. Science and logic seem sufficient to tell me whether, for instance, a utilitarian standard of ethics satisfies certain goals that many of us would agree on. I don’t need religion to set the goals, and I don’t need religion to tell me how to achieve the goals that I’ve set.

Now, Christianity probably doesn’t need to spend any time convincing me. It has a billion or so people who are perfectly willing to believe in its tenets. But is it really the case that every single one of those people is a non-scientist? Or do scientists somehow keep their religion in a separate compartment? Are there any good books to explain the allure of Christianity to those with a scientific temperament?

The Sox’ baseball for the last couple days has been really, really ugly. This is just not fun to watch. It’s one thing to lose even when you’re trying your best. It’s quite another to lose in large part because your team keeps making ugly errors.

P.S.: The Sox were pulling it out, and then Keith Foulke had a meltdown. He just allowed a grand slam, turning what had been an 8-5 Sox lead (as of the time when Timlin took the mound, I believe) into a 12-8 Indians lead. If the Sox pull this out in the bottom of the 9th, it will be a miracle.

In fairness, the final out (or at least the second — I can’t remember) would have come much earlier, but the ball slipped out of Bellhorn’s hand on a tricky backhand tag at second. Still, Foulke wasn’t exactly exemplary from that point on.

Everyone’s writing about the fallout from the Grokster case. Lots of good stuff out there; just google for Grokster and let your fingers do the walking. (Incidentally, it’s sad that our children won’t get that reference. I also lament the demise of Max Headroom knowledge, and the end of Where’s The Beef jokes.)

One of my big questions is whether this decision provides any practical guidance to future technologists, and whether this will clarify the legal situation to such an extent that a lot of cases will end before they arrive at the courts. The next big technology to land in court will probably be BitTorrent, which the recording industries hate. So it’s interesting to see Ed Felten’s take on the future of BitTorrent. He can’t imagine that the recording industries will stand for its continued existence, even though to all appearances it’s a non-infringing technology even by the new “inducement” standard.

I every now and again see people quoting shell commands to me that contain awk. Is there any good reason to use awk, given that I know Perl and sed and a whole bunch of other little command-line tools that are basically guaranteed to be on a user’s machine? (E.g., cut, sort, uniq, echo, grep, tr, rename, etc., etc.)

I recently started using Akregator as my newsreader. It’s good stuff, and definitely better than Straw. It’s probably the Linux newsreader that most closely resembles SharpReader, which is the best Windows newsreader that I’ve used.

Still nothing quite matches NetNewsWire, but that may have more to do with OS X than with NetNewsWire specifically.

Again with the same client’s site, I’m trying to use a bare minimum of architecture to get some templating, because they really don’t need a database backend or any kind of scripting — they need basically just a teensy bit more than what SSI gives them.

The logic I need is this: if a file called ${QUERYSTRING}masthead.html exists, include it. Otherwise, if a file called masthead.html exists, include it. If neither of those files exist, check for a file called /masthead.html and include it.

The trouble is that it doesn’t look like SSI contains enough meat to tell me whether a file exists. Which is kind of odd, because if you do

<!--#include virtual="filename.html" -->

and filename.html doesn’t exist, SSI will give you an error. So obviously it has enough architecture built in to know whether the file is there. Shouldn’t it have enough architecture to tell me this fact?

Here’s a challenge. A client of mine, back in the day, couldn’t set up much server-side architecture, so they ended up pushing a lot of their website into every single page. So every page, for instance, contains an instruction to execute a bit of Javascript code. That bit of Javascript code’s sole purpose is to read in a bit of boilerplate text — normally a small menu that changes every few weeks. So every page ends up looking 90% identical to every other page, with the exception of whatever unique stuff is in the center column; the masthead, sidebar, footer, navigation menus and so forth are all basically identical from page to page. The only mildly complicated bit is that my client has a number of different projects, and each project gets slightly customized bits — say, a different masthead.  . . .  Actually, I think that’s the only customized bit.

Now they’ve just asked me to replace a lot of their graphical menus with text-based ones, which will involve doing a massive search-and-replace across probably a few thousand HTML files. It shouldn’t be a big deal, given grep and sed and perl and a few other tools, but certainly I don’t want to be doing this sort of thing for them for years. I’d much rather expend the effort now to set them up with some templates, and save myself much monotony in the future.

But let’s suppose for a moment that I can’t do anything on the server side for them — that their server effectively only allows HTML. (I suspect that it’s unreasonable to believe this; most ISPs at least allow SSI, which can do a lot in concert with mod_rewrite.) Is there any way to do client-side templating? I push a template out to the browser, then the browser downloads all the parts — masthead, sidebar, etc. — that it needs. Voilà: no server-side template, but also no monotonous repetition of HTML across many pages.

Possible?

Crap: the Court has ruled unanimously against Grokster. That’s bad. I wonder whether William Patry’s prediction will turn out to be altogether too optimistic.

Then again, perhaps we should keep Fred von Lohmann’s advice in mind.

P.S.: Everybody has something to say about this. I think Eric Goldman captures most of it.

I’ve not read the decision yet (though I have cached everything for later reading). What I’m concerned about is the “inventor in the garage”: will the inventor of the next iPod have to somehow ensure that his technology is unusable for infringing purposes? Will he, for instance, have to build in some sort of filtering technology to detect the presence of copyrighted works? My sense is that the decision doesn’t say such a thing, and that it focuses fairly narrowly on Grokster’s obvious intent to distribute copyrighted works. Future inventors who don’t brazenly seek to violate the law will not be prosecuted.

But then I’ve not read the decision. I’m hopeful, though.

Wow. That’s awesome. If I had $8,000 to spare, I would totally snap that up.

(Via Slashdot)

Reading a New York Times article about Peter Jackson’s lawsuit against New Line, an economics question that I’ve never satisfactorily resolved has popped back into my head. The question is this: since Jackson is suing for $100 million, why wouldn’t the studio spend up to $99,999,999.99 to defend themselves? Yet the traditional economic analysis — and the one that Richard Posner uses throughout his law-and-economics work — says that the studio should add one dollar to its expenditures precisely up to the point that they get less than one dollar of benefit back. That is, they’re basing their decisions on the marginal benefit of their spending, rather than the aggregate benefit.

Can someone explain to me why that is?  . . .  Oh. I think I figured it out, as soon as I wrote it down above. Suppose that $50 million is the optimal amount, in the sense that dollars spent above $50 million yield less than $1 in benefit. Suppose that the marginal benefit is 90 cents for every dollar spent between $50 million and $100 million, and that the benefit is $1.01 for every dollar below $50 million. Then if the law firm spends $100 million, they’re getting back 1.01($50 million) + .9($50 million) = $95.5 million in benefit on $100 million in expense. Whereas if they spent $50 million, their total benefit would be $50.5 million. If the production company is assumed to be profit-maximizing, it would never spend more than $50 million on legal fees. Indeed, it would never spend less than $50 million, because by adding additional dollars up to $50 million it could increase its profit. So it will always spend precisely $50 million.

The organization would continue to spend marginal dollars above $100 million if and only if they were getting back more than $1 in benefit from those marginal dollars. But I can’t quite imagine how that could be the case, given that the total amount of money at stake to begin with is only $100 million. So it seems like it must be the case that the marginal return becomes less than the marginal expense somewhere below $100 million. Does the theory predict that?

I think that’s basically the argument. If anyone can add detail, feel free to do so in the comments.

Wow: this is just insanely cool:

The New Yorker is selling a limited edition set of 8 DVDs containing every page of the magazine from its inception in February 1925 to February 2005: ‘from full-color covers to spot drawings, from poetry to Profiles, from cartoons to advertisements — on reader friendly and highly searchable DVDs.’ It’ll be available in September, and will run on Windows and Macs.

I mean, except for the just-on-Windows-and-Mac bit. But I bet it’ll work under Linux using Wine. Awesome.

Willard van Orman Quine, Mathematical Logic, Revised Edition, 1958, p. 13 footnote 1:

The series of adjectives ‘binary’, ‘ternary’, ‘quaternary’, ‘quinary’,  . . .  leaves mathematicians in a quandary when n = 1. It is customary to stammer out some such makeshift as ‘unary’ or ‘uninary’ or ‘unitary’. But the proper word is apparent if we reflect that the series of Latin distributives ‘bini’, ‘terni’, ‘quaterni’, ‘quini’,  . . .  begins with ‘singuli’.

Latin and Quine to the contrary, usage seems to have adopted “unary.” Hence the operator prepended to a positive integer to give it a negative sign is known as “unary minus” in any computer-science textbook that you care to consult; I have never seen “singulary” in any computer-science context, though it looks like at least a few people use it (even excluding the typos that were obviously meant to be “singularly”) in other contexts.