Older blog entries for chalst (starting at number 101)

AI and the semantic web
There's an interesting discussion going on in various places about the goals of research on universal metadata and the semantic web, turning on the questions: is it all too ambitious? Is it stupid? First, some references:

1. Clay Shirky says that the semantic web is based on a lie, namely you can do interesting things with syllogisms;
2. Paul Ford writes a wonderful counterargument to Shirky, saying he attacks a straw man, and that it is all about doing the same kind of thing that librarians do with bibliographic data;
3. Mark Liberman tells a nice story of an NSF grant proposal; ontologies and arguments.
4. Poor but happy Peter Van Dijck has a nice summary of the key ideas at work in the semantic web discussion.
5. And now, Mark Liberman reports on a dreadful conspiracy by the AI gnomes of Zurich. Choice quote:

   As for the content, I don't think anyone (in this discussion) will admit to being opposed to vision. The thing is, some visions turn out to be the telephone or the automobile or the internet, while others turn out to be the Picturephone, the Personal Zeppelin[TM] or perhaps the Philosophical Language of John Wilkins. This is an argument for pluralism but against credulousness.

A point about the vision thing: it really isn't clear that in the long term we need web standards. Why can't we have a standards-free commons of mob software quick-hackers with nice, practical AI agents figuring out what is meant by all of the ad-hoc order that they find? To take this further: we need html and http; we have XML, XHTML, CSS, RDF, RSS, and that's nice, but do we really need them? It seems to me the only thing that matters with these semantic web issues is the forseeable future, and we can't forsee very far.

There's some consequences that follow from this, but I have to work, so it will have to wait.

Utility and Reference
There's been an interesting discussion about act utilitarianism and what Larry Solum calls the Demandingness Objection. A couple of people have pointed out correctly that Richard Hare's 2-level utilitarianism is immune to this objection.

I'd like to draw attention to the fact, for those who follow analytical philosophy, that there is quite a close analogy between the way Hare achieves this and the structure of Frege's theory of Sinn and Bedeutung (esp. as articulated by Michael Dummett). Essentially the relation of sense to reference is somewhat analogous to the relation between Hare's common-sense obligation and utility, and it is not difficult to make the analogy very strong by means of something such as a posited notion of common-sense moral worth. Michael Dummett says of sense that it is the normal point of utterances (which have sense) to refer, but they need not be successful; likewise with the relation between common-sense moral worth and utility.

References
[1] Larry Solum's original post.
[2] Larry Solum's later comment.
[3] Ciceronian Review's post. This new blog is well worth watching.

Recentlog comments
salmoni: Bloody hell!
dyork: I'm iCAN, and I don't listen to the Archers. I try again, with slightly different answers and I am USENET, but I don't even remember alt.lemur.frink.frink.frink...
cerquide: Got to say, I had a much, much more positive experience with my doctorate than in Peter Burney's description. My experience of industry wasn't fractionally as good by comparison.

Rumblings in the Blogosphere
Interesting! Via the Truth Laid Bear, I see that Attention Deficit has overtaken Instapundit as most linked to weblog. Soul-laid-bare single displaces elite warblogger. TTLB is an hourly updated service; I wonder if this will last...
Update No, it was just a spike.

Which member of the Bush administration are you?
I had no idea, but it turns out that, like Stephen Pollard

I'm George Walker Bush! I'm the most powerful man in the world! I have little time to think for myself, but fortunately, I have my friends to think for me...

The powers behind the throne are: Norman Geras as Condi, Damian Penny as Dick Cheney, diablod3 as Colin Powell, zinger as Karl Rove, Horologium as John Ashcroft, and The Country Pundit as Donald Rumsfeld. Which member are you?

Recommendations Sought
raph: Ah! You like Talking Points Memo too. Any recommendations in my other categories?

Workshops and Blogging
We've just finished an excellent two day workshop here in Dresden, with participants mostly local and from France. If some kind of theme emerged from the workshop, I would say it was the need to understand powerful ideas in geometry to help proof theory advance to a healthier condition, a claim forcefully and effectively articulated by Francois Lamarche.

I want to write up my reflections on the workshop, filling in these details, which I hope will follow in the next few days. Larry Solum has some important reflections on conference blogging, which I think are worth wide consideration. I'm talking to Larry about the issues he raises; my main point is I think the value of conference blogging far outweighs its costs, a point on which we are in agreement and perhaps Brian Leiter is not.

Confessions of a Political Nature
There is an interesting confession by someone confused as to whether they are liberal or conservative.

Did the USA Betray Iraq Twelve Years Ago?
Some reflections from Dan Drezner.

Across the Abyss
Raph appears to be despairing of the general political debate between left&right. From one of his recent messages:

I'm having a number of fascinating correspondences in email now. For one, I decided to follow up by email to Berend De Boer's recommendation of a piece criticizing a recent op-ed by Krugman, and I'm glad of it. I think there's a good chance we'll both learn something from the discourse -- something that seems to be happening more rarely as the public debate between "liberals" and "conservatives" seems to be getting more polarized.

Raph's proposition, that the public debate between left and right seems to be getting more polarized, I think this is true of the US at a high-level, but not true at a lower level, and perhaps this lower level is more important. Let me make some points:

1. I think it's mostly a US phenomena, and is driven by a process that disgracefully the US media draws little attention to, namely the cosy agreement that exists between Democrats and Republicans to `jerrymander' electoral boundaries, to ensure that marginal Democrat and Republican seats become safe seats. This means fewer politicians need work to please centrists, and advancement becomes principally a matter of pleasing the party base, who tend to be loudest at the fringes.
2. As a general rule, the polarisation of representative politics leads to a polarisation of political discussion in the mainstream media.
3. Also consider 2-party (ie. first past the post), vs. pluralist systems. I think avoiding extremism needs consitutional measures (in the widest sense of that term), in just the same way as the need to eliminate totalitarianism in many European required consitutional measures at the end of WWII.
4. The advent of blogging is a good thing:
   • There are rightists respected by many leftists, eg. Arnold Kling, Dan Drezner, Eugene Volokh.
   • There are leftists respected by many rightists, eg. Kevin Drum, Josh Marshall, and Norman Geras.
   • There are also centrists who can talk to both left and right; eg. Michael J. Totten. Oliver Kamm, although he calls himself a leftist, perhaps fits best in this category.
   For people who think, the interest of ideas is more important than the conformity of their authors. That the blogger phenomenon has sparked off a global conversation where many can easily participate and noone can shout down anyone else is the important sense in which raph's proposition is false.
5. This goes alongside an important phenomenon that has been going on over the last 20-30 years, firstly on the right (since Barry Goldwater, and as a mass phenomenon since Reagan), and is now gathering steam on the left (Clinton/Blair), namely a fluidity of ideas between left and right. There aren't really any single propositions that mark someone as being a leftist or a rightist anymore; in a sense we are leaving ideology behind. Again, the right is much more advanced in this respect than the left, and I think this is why the left tends to be, to cite George Lakoff, much worse at "framing problems" than the right. None of this is to say that the right is not filled with quite as many dogmatic idiots as the left...
6. In the above points, when I talk about left and right, I mean those who accept the idea of a state based on principles of tolerance and the rule of law. Communists, ultranationalists and religious fundamentalists are beyond left and right in this sense.

I've more to say on this, but it'll have to wait.

Luskin sleaze watch III
Dan Drezner calls Luskin a stalker.

Worst album covers ever
Via Gene at Harry's place, enjoy this feast for the eyes.

What motivates good science
Randy Barnett argues that the technological spin offs that come from military spending are unlikely to justify the huge sums involved: read his thoughts on Star Trek and Democracy, and on War and Technological Development.

I think he's probably right, but I didn't think the argument he gives is sound. Two considerations:

• War does effectively separate efficient states from failing states. I hope we don't need it, but the fact that Europe had near perpetual wars for 6 of the last 7 centuries I think is how Europe came to overtake its then far superior rivals in development terms (the Islamic world, China, India, etc.). I'd like to believe that trade favours the selection of efficient states over failing states more efficiently in the long term, since it is less horrible than war, but I'm not yet completely convinced.
• Fundamental science does not advance in step with technological improvements. Markets promote the latter well, I am not impressed by their ability to promote the former. The non-market driven model of government-funded research foundations (eg. in the US, the NIH and the NSF, in the UK the SERC etc., and Europe-wide the Marie-Curie framework programmes) do a much better job. The obvious reason is that no better way exists of telling what ideas are worth pursuing than the opinion of good people working in the domain of enquiry, and no better way exists way of telling who is good than letting the test of time sort the people with good ideas from those with bad ideas.

Relative Consistency
I was surprised and delighted to discover that Robert Solovay, a logician whom I admire, reads my diary. He's picked up an important issue with my post promoting Proofs and Types, namely that consistency as defined by Goedel, the inability to prove 0=1 is too weak a notion; 1-consistency is the right notion. I have more to say about this, which I will post in a future diary entry.

Robert also made an interesting comment on a post from two months ago on Bram's Minefield game. I said then that:

In Bram's 6th August entry Bram introduces a `game' played over ZFC: even in the simplified case of large cardinal axioms (LCAs), there may be no fact of the matter of which is stronger: if you look at the diagram of important LCAs in Kanamori's "The Higher Infinite", they form a partial order, not a linear order (I don't know whether this reflects known independence results or just ignorance). I do think as a kind of thought experiment, your exercise is illuminating; I like to think about how different mathematical pholosophy might be if results came to known in different order, might we regard a much weaker or stronger set theory as the default?

Robert replied that:

I don't have Kanamori's table in front of me. But the general consensus is that LCA's are linearly ordered by consistency strength in the following sense:

If A and B are LCA's then either 1) PA proves "Con(ZFC + A) iff Con(ZFC + B)" or 2) ZFC + A proves Con (ZFC + B) or 3) ZFC + B proves Con(ZFC + A).

Some comments:

1. There are two "base theories" in Robert's remark. The first is Peano Arithmetic (PA), which can be replaced by any theory that is strong enough to `talk' about consistency. Primitive Recursive Arithmetic (PRA) is a good theory, weaker than PA, interesting especially because it is logic-free: it consists solely of equations between arithmetic expressions. There are weaker theories than PRA, and interesting theories of arithmetic that are too weak, for example Robinson's theory Q, sometimes called Robinson Arithmetic.
2. The second is Zermelo-Fraenkel set theory (with choice). We can reasonably omit the axiom of infinity from our base theory, getting a well-studied finitary system. We could instead omit the power set axiom, in which we would get a system rather like Z_2 (second-order arithmetic). In each case we can play Bram's game over the new base theory. Question: for the first alternate base theory, obviously the game is almost the same; is this so for the second (ie. without the power set axiom)>
3. If the consensus is true, then one can image that what much of the set-theoretic community is doing is exactly playing Bram's game. My guess is, for what little it is worth, that if we make reasonably restrictions about the form that LCAs can take, then the consensus view is false.
4. It seems appropriate to report a recent (two years old) breakthrough in the theory of Goedelian incompleteness. I talked above about base theories that talk about completeness. Most theories fall into one of two categories: either they are too weak to talk about consistency, or they prove that they cannot prove their own consistency. A natural question arises: are there theories strong enough to talk about consistency that are weak enough to prove their own consistency? Most people who understand the question would say no; it turns out that in fact there are such theories, as proven by Dan Willard.

Explanations and mechanical proofs
There is an interesting discussion on the FOM list regarding what is an explanatory proof. Harvey Friedman connects this to proof assistants in a shocked message.

Update
Robert Solovay points out another error in the above...

ncm: Why do you think raph was "taken in" by Econopundit? He's only said berend's recommendation led to an interesting discussion.

What Liberal Media?
I've it mentioned it briefly before, but it seems to be becoming more topical, so here goes: whilst it is a staple of the American right that news sources such as the New York Times, the Washington Post and CNN have a liberal bias, there's a rather effective argument claiming the contrary, namely there is a subtle but influential right wing bias in these papers. To summarise:

• Eric Alterman effectively makes this argument in his book What liberal media?, which has a sizable excerpt at The Nation.
• The claim is not that there is no liberal media, publications such as The Nation are most certainly liberal, and papers such as The New Republic and The New York Review of Books have a perceptible liberal bias. The claim is that the right perceive left-wing bias in the majority of mainstream publications where there is none.
• That reporters in general tend to be left leaning is a well documented fact (eg. in surveys of Democrat vs. Republican voting habits, among journalists at big newspapers). It in no way contradicts the basic thesis, a claim that many right-wing media commentators seem to find hard to accept.
• Though not accepting the thesis, several right-wing pundits accept that the media is becoming increasingly friendly to right wing views. Daniel Drezner provides a good commentary.
• Dean Baker's Economic Reporting Review documents what he sees as misleading reporting of economic stories by the New York Times and Washington Post. Together his stories provide a strong case that these papers systematically favour right-wing economic viewpoints over left-wing viewpoints, and favour the prespectives of capital markets and executives of Fortune 500 companies over perspectives of labour organisations and consumers.

Luskin Sleaze Squad
Calpundit has written a condemnation of Donald Luskin's recent edition of Krugman Truth Squad in which he cheaply accuses Krugman of anti- semitism; Brad DeLong also has some nice irony. I never had a high opinion of Luskin, now I think he's a maggot. Will berend still feel he should defend this guy? And Econopundit?

New Weblog
Take a look at The Importance Of.

Update: solovay pointed out a typo: TNR is perhaps liberal; the National Review certainly isn't.

On Krugman
Krugmanology seems to be becoming a more and more popular sport these days... It's nice to see that berend and raph are having a discussion of Krugman's book: I'd be interested to hear what comes of it. Thanks to berend for giving me the link to Econopundit (Steve Antler). A quick thought, after reading Oliver Kamm on Krugman's pathology: I think most critics of Krugman miss the point, namely the important thesis of Krugman's is whether or not Bush is pushing through a revolutionary agenda that is being ignored by the mainstream media (subthesis: due to the Bush administration providing an endlessly changing stream of rationales for an unchanging policy), and that criticism of Krugman has to be centred on whether or not this thesis is true. The usually astute Oliver Kamm seems to have missed the point here.

For more sympathetic Krugmanology, see the current New York Review of Books on The Awful Truth. Happy 40th Anniversary to the NYRB, btw; they've put everything online so we can celebrate.

Oil
While I mention berend, with regards to his recent prediction about oil, his argument is supported by the recent issue of the Economist, which argues we should end our oil dependency. Whether you agree with their argument or not, it seems clear that the alternatives to oil are now practicable, even if still somewhat expensive.

News
Well, not so new news; I guess just about everyone who reads this diary knows already, but better late than never: my wife is six months pregnant; our baby daughter is due in January. Life will change...

berend: I think a credible hatchet job on the economic case in Krugman's book needs to be done by someone competent in economics, and I don't think Donald Luskin is much of an authority (FWIW, Luskin has something of a bee in his bonnet about DeLong. Luskin's response to DeLong's comment is here).

A couple of links
In the discussion about appropriate consequence relations for doing proof- checking that was discussed here almost a year ago, raph asked me if I could recommend an online source to the theory of the polymorphic lambda-calculus. I couldn't, but I suggested the Girard-Lafont-Taylor book as the best offline reference. Some news: the book has gone out of print, and Paul Taylor has made the full text of Proofs and Types available online. So now no-one has any excuse for not knowing why strong normalisation of the polymorphic lambda calculus is equivalent to the 1-consistency of second-order arithmetic...

Also available online, is a tutorial by two of my colleagues, Alessio Guglielmi and Paola Bruscoli, on the Proof Theoretic Foundations of Logic Programming.

Update: Following a comment by solovay, I've tightened the above statement relating Z2 to system F. There's more in this diary entry.