Category Archives: Uncategorized

Room to grow

However, Apple CEO Tim Cook pointed out that 60% of people who owned an iPhone before the launch of the iPhone 6 haven’t upgraded to the most recent models, which implies that there is still room to grow, Reuters notes.

Doesn’t it imply that a) people are no longer on contracts incentivizing biannual upgrade; and b) Apple hasn’t figured out a way to make a new phone that’s different enough from the iPhone 5 to make people want to switch?

Messing around with word2vec

Word2vec is a way of representing words and phrases as vectors in medium-dimensional space developed by Tomas Mikolov and his team at Google; you can train it on any corpus you like (see Ben Schmidt’s blog for some great examples) but the version of the embedding you can download was trained on about 100 billion words of Google News, and encodes words as unit vectors in 300-dimensional space.

What really got people’s attention, when this came out, was word2vec’s ability to linearize analogies.  For example:  if x is the vector representing “king,” and y the vector representing “woman,” and z the vector representing “man,” then consider

x + y – z

which you might think of, in semantic space, as being the concept “king” to which “woman” has been added and “man” subtracted — in other words, “king made more female.”  What word lies closest in direction to x+y-z?  Just as you might hope, the answer is “queen.”

I found this really startling.  Does it mean that there’s some hidden linear structure in the space of words?

It turns out it’s not quite that simple.  I played around with word2vec a bunch, using Radim Řehůřek’s gensim package that nicely pulls everything into python; here’s what I learned about what the embedding is and isn’t telling you.

Word2Vec distance isn’t semantic distance

The Word2Vec metric tends to place two words close to each other if they occur in similar contexts — that is, w and w’ are close to each other if the words that tend to show up near w also tend to show up near w’  (This is probably an oversimplification, but see this paper of Levy and Goldberg for a more precise formulation.)  If two words are very close to synonymous, you’d expect them to show up in similar contexts, and indeed synonymous words tend to be close:

>>> model.similarity(‘tremendous’,’enormous’)

0.74432902555062841

The notion of similarity used here is just cosine distance (which is to say, dot product of vectors.)  It’s positive when the words are close to each other, negative when the words are far.  For two completely random words, the similarity is pretty close to 0.

On the other hand:

>>> model.similarity(‘tremendous’,’negligible’)

0.37869063705009987

Tremendous and negligible are very far apart semantically; but both words are likely to occur in contexts where we’re talking about size, and using long, Latinate words.  ‘Negligible’ is actually one of the 500 words closest to ’tremendous’ in the whole 3m-word database.

You might ask:  well, what words in Word2Vec are farthest from “tremendous?”  You just get trash:

>>> model.most_similar(negative=[‘tremendous’])

[(u’By_DENISE_DICK’, 0.2792186141014099), (u’NAVARRE_CORPORATION’, 0.26894450187683105), (u’By_SEAN_BARRON’, 0.26745346188545227), (u’LEGAL_NOTICES’, 0.25829464197158813), (u’Ky.Busch_##-###’, 0.2564955949783325), (u’desultorily’, 0.2563159763813019), (u’M.Kenseth_###-###’, 0.2562236189842224), (u’J.McMurray_###-###’, 0.25608277320861816), (u’D.Earnhardt_Jr._###-###’, 0.2547803819179535), (u’david.brett_@_thomsonreuters.com’, 0.2520599961280823)]

If 3 million words were distributed randomly in the unit ball in R^300, you’d expect the farthest one from “tremendous” to have dot product about -0.3 from it.  So when you see a list whose largest score is around that size, you should think “there’s no structure there, this is just noise.”

Antonyms

Challenge problem:  Is there a way to accurately generate antonyms using the word2vec embedding?  That seems to me the sort of thing the embedding is not capturing.  Kyle McDonald had a nice go at this, but I think the lesson of his experiment is that asking word2vec to find analogies of the form “word is to antonym as happy is to?” is just going to generate a list of neighbors of “happy.”  McDonald’s results also cast some light on the structure of word2vec analogies:  for instance, he finds that

waste is to economise as happy is to chuffed

First of all, “chuffed” is a synonym of happy, not an antonym.  But more importantly:  The reason “chuffed” is there is because it’s a way that British people say “happy,” just as “economise” is a way British people spell “economize.”  Change the spelling and you get

waste is to economize as happy is to glad

Non-semantic properties of words matter to word2vec.  They matter a lot.  Which brings us to diction.

Word2Vec distance keeps track of diction

Lots of non-semantic stuff is going on in natural language.  Like diction, which can be high or low, formal or informal, flowery or concrete.    Look at the nearest neighbors of “pugnacity”:

>>> model.most_similar(‘pugnacity’)

[(u’pugnaciousness’, 0.6015268564224243), (u’wonkishness’, 0.6014434099197388), (u’pugnacious’, 0.5877301692962646), (u’eloquence’, 0.5875781774520874), (u’sang_froid’, 0.5873805284500122), (u’truculence’, 0.5838015079498291), (u’pithiness’, 0.5773230195045471), (u’irascibility’, 0.5772287845611572), (u’hotheadedness’, 0.5741063356399536), (u’sangfroid’, 0.5715578198432922)]

Some of these are close semantically to pugnacity, but others, like “wonkishness,” “eloquence”, and “sangfroid,” are really just the kind of elevated-diction words the kind of person who says “pugnacity” would also say.

In the other direction:

>>> model.most_similar(‘psyched’)

[(u’geeked’, 0.7244787216186523), (u’excited’, 0.6678282022476196), (u’jazzed’, 0.666187584400177), (u’bummed’, 0.662735104560852), (u’amped’, 0.6473385691642761), (u’pysched’, 0.6245862245559692), (u’exicted’, 0.6116108894348145), (u’awesome’, 0.5838013887405396), (u’enthused’, 0.581687331199646), (u’kinda_bummed’, 0.5701783299446106)]

“geeked” is a pretty good synonym, but “bummed” is an antonym.  You may also note that contexts where “psyched” is common are also fertile ground for “pysched.”  This leads me to one of my favorite classes of examples:

Misspelling analogies

Which words are closest to “teh”?

>>> model.most_similar(‘teh’)

[(u’ther’, 0.6910992860794067), (u’hte’, 0.6501408815383911), (u’fo’, 0.6458913683891296), (u’tha’, 0.6098173260688782), (u’te’, 0.6042138934135437), (u’ot’, 0.595798909664154), (u’thats’, 0.595078706741333), (u’od’, 0.5908242464065552), (u’tho’, 0.58894944190979), (u’oa’, 0.5846965312957764)]

Makes sense:  the contexts where “teh” is common are those contexts where a lot of words are misspelled.

Using the “analogy” gadget, we can ask; which word is to “because” as “teh” is to “the”?

>>> model.most_similar(positive=[‘because’,’teh’],negative=[‘the’])

[(u’becuase’, 0.6815075278282166), (u’becasue’, 0.6744950413703918), (u’cuz’, 0.6165347099304199), (u’becuz’, 0.6027254462242126), (u’coz’, 0.580410361289978), (u’b_c’, 0.5737690925598145), (u’tho’, 0.5647958517074585), (u’beacause’, 0.5630674362182617), (u’thats’, 0.5605655908584595), (u’lol’, 0.5597798228263855)]

Or “like”?

>>> model.most_similar(positive=[‘like’,’teh’],negative=[‘the’])

[(u’liek’, 0.678846001625061), (u’ok’, 0.6136218309402466), (u’hahah’, 0.5887773633003235), (u’lke’, 0.5840467214584351), (u’probly’, 0.5819578170776367), (u’lol’, 0.5802655816078186), (u’becuz’, 0.5771245956420898), (u’wierd’, 0.5759704113006592), (u’dunno’, 0.5709049701690674), (u’tho’, 0.565370500087738)]

Note that this doesn’t always work:

>>> model.most_similar(positive=[‘should’,’teh’],negative=[‘the’])

[(u’wil’, 0.63351970911026), (u’cant’, 0.6080706715583801), (u’wont’, 0.5967696309089661), (u’dont’, 0.5911301970481873), (u’shold’, 0.5908039212226868), (u’shoud’, 0.5776053667068481), (u’shoudl’, 0.5491836071014404), (u”would’nt”, 0.5474458932876587), (u’shld’, 0.5443994402885437), (u’wouldnt’, 0.5413904190063477)]

What are word2vec analogies?

Now let’s come back to the more philosophical question.  Should the effectiveness of word2vec at solving analogy problems make us think that the space of words really has linear structure?

I don’t think so.  Again, I learned something important from the work of Levy and Goldberg.  When word2vec wants to find the word w which is to x as y is to z, it is trying to find w maximizing the dot product

w . (x + y – z)

But this is the same thing as maximizing

w.x + w.y – w.z.

In other words, what word2vec is really doing is saying

“Show me words which are similar to x and y but are dissimilar to z.”

This notion makes sense applied any notion of similarity, whether or not it has anything to do with embedding in a vector space.  For example, Levy and Goldberg experiment with minimizing

log(w.x) + log(w.y) – log(w.z)

instead, and get somewhat superior results on the analogy task.  Optimizing this objective has nothing to do with the linear combination x+y-z.

None of which is to deny that the analogy engine in word2vec works well in many interesting cases!  It has no trouble, for instance, figuring out that Baltimore is to Maryland as Milwaukee is to Wisconsin.  More often than not, the Milwaukee of state X correctly returns the largest city in state X.  And sometimes, when it doesn’t, it gives the right answer anyway:  for instance, the Milwaukee of Ohio is Cleveland, a much better answer than Ohio’s largest city (Columbus — you knew that, right?)  The Milwaukee of Virginia, according to word2vec, is Charlottesville, which seems clearly wrong.  But maybe that’s OK — maybe there really isn’t a Milwaukee of Virginia.  One feels Richmond is a better guess than Charlottesville, but it scores notably lower.  (Note:  Word2Vec’s database doesn’t have Virginia_Beach, the largest city in Virginia.  That one I didn’t know.)

Another interesting case:  what is to state X as Gainesville is to Florida?  The answer should be “the location of the, or at least a, flagship state university, which isn’t the capital or even a major city of the state,” when such a city exists.  But this doesn’t seem to be something word2vec is good at finding.  The Gainesville of Virginia is Charlottesville, as it should be.  But the Gainesville of Georgia is Newnan.  Newnan?  Well, it turns out there’s a Newnan, Georgia, and there’s also a Newnan’s Lake in Gainesville, FL; I think that’s what’s driving the response.  That, and the fact that “Athens”, the right answer, is contextually separated from Georgia by the existence of Athens, Greece.

The Gainesville of Tennessee is Cookeville, though Knoxville, the right answer, comes a close second.

Why?  You can check that Knoxville, according to word2vec, is much closer to Gainesville than Cookeville is.

>>> model.similarity(‘Cookeville’,’Gainesville’)

0.5457580604439547

>>> model.similarity(‘Knoxville’,’Gainesville’)

0.64010456774402158

But Knoxville is placed much closer to Florida!

>>> model.similarity(‘Cookeville’,’Florida’)

0.2044376252927515

>>> model.similarity(‘Knoxville’,’Florida’)

0.36523836770416895

Remember:  what word2vec is really optimizing for here is “words which are close to Gainesville and close to Tennessee, and which are not close to Florida.”  And here you see that phenomenon very clearly.  I don’t think the semantic relationship between ‘Gainesville’ and ‘Florida’ is something word2vec is really capturing.  Similarly:  the Gainesville of Illinois is Edwardsville (though Champaign, Champaign_Urbana, and Urbana are all top 5) and the Gainesville of Indiana is Connersville.  (The top 5 for Indiana are all cities ending in “ville” — is the phonetic similarity playing some role?)

Just for fun, here’s a scatterplot of the 1000 nearest neighbors of ‘Gainesville’, with their similarity to ‘Gainesville’ (x-axis) plotted against their similarity to ‘Tennessee’ (y-axis):

Screen Shot 2016-01-14 at 14 Jan 4.37.PM

The Pareto frontier consists of “Tennessee” (that’s the one whose similarity to “Tennessee” is 1, obviously..) Knoxville, Jacksonville, and Tallahassee.

Bag of contexts

One popular simple linear model of word space is given by representing a word as a “bag of contexts” — perhaps there are several thousand contexts, and each word is given by a sparse vector in the space spanned by contexts:  coefficient 0 if the word is not in that context, 1 if it is.  In that setting, certain kinds of analogies would be linearized and certain kinds would not.  If “major city” is a context, then “Houston” and “Dallas” might have vectors that looked like “Texas” with the coodinate of “major city” flipped from 0 to 1.  Ditto, “Milwaukee” would be “Wisconsin” with the same basis vector added.  So

“Texas” + “Milwaukee” – “Wisconsin”

would be pretty close, in that space, to “Houston” and “Dallas.”

On the other hand, it’s not so easy to see what relations antonyms would have in that space. That’s the kind of relationship the bag of contexts may not make linear.

The word2vec space is only 300-dimensional, and the vectors aren’t sparse at all.  But maybe we should think of it as a random low-dimensional projection of a bag-of-contexts embedding!  By the Johnson-Lindenstrauss lemma, a 300-dimensional projection is plenty big enough to preserve the distances between 3 million points with a small distortion factor; and of course it preserves all linear relationships on the nose.

Perhaps this point of view gives some insight into which kind of word relationships manifest as linear relationships in word2vec.  “flock:birds” is an interesting example.  If you imagine “group of things” as a context, you can maybe imagine word2vec picking this up.  But actually, it doesn’t do well:

>> model.most_similar(positive=[‘fish’,’flock’],negative=[‘birds’])
[(u’crays’, 0.4601619839668274), (u’threadfin_salmon’, 0.4553075134754181), (u’spear_fishers’, 0.44864755868911743), (u’slab_crappies’, 0.4483765661716461), (u’flocked’, 0.44473177194595337), (u’Siltcoos_Lake’, 0.4429660737514496), (u’flounder’, 0.4414420425891876), (u’catfish’, 0.4413948059082031), (u’yellowtail_snappers’, 0.4410281181335449), (u’sockeyes’, 0.4395104944705963)]

>> model.most_similar(positive=[‘dogs’,’flock’],negative=[‘birds’])
[(u’dog’, 0.5390862226486206), (u’pooches’, 0.5000904202461243), (u’Eminem_Darth_Vader’, 0.48777419328689575), (u’Labrador_Retrievers’, 0.4792211949825287), (u’canines’, 0.4766522943973541), (u’barked_incessantly’, 0.4709487557411194), (u’Rottweilers_pit_bulls’, 0.4708423614501953), (u’labradoodles’, 0.47032350301742554), (u’rottweilers’, 0.46935927867889404), (u’forbidding_trespassers’, 0.4649636149406433)]

The answers “school” and “pack” don’t appear here.  Part of this, of course, is that “flock,” “school”, and “pack” all have interfering alternate meanings.  But part of it is that the analogy really rests on information about contexts in which the words “flock” and “birds” both appear.  In particular, in a short text window featuring both words, you are going to see a huge spike of “of” appearing right after flock and right before birds.  A statement of the form “flock is to birds as X is to Y” can’t be true unless “X of Y” actually shows up in the corpus a lot.

Challenge problem:  Can you make word2vec do a good job with relations like “flock:birds”?  As I said above, I wouldn’t have been shocked if this had actually worked out of the box, so maybe there’s some minor tweak that makes it work.

Boys’ names, girls’ names

Back to gender-flipping.  What’s the “male version” of the name “Jennifer”?

There are various ways one can do this.  If you use the analogy engine from word2vec, finding the closest word to “Jennifer” + “he” – “she”, you get as your top 5:

David, Jason, Brian, Kevin, Chris

>>> model.most_similar(positive=[‘Jennifer’,’he’],negative=[‘she’])
[(u’David’, 0.6693146228790283), (u’Jason’, 0.6635637283325195), (u’Brian’, 0.6586753129959106), (u’Kevin’, 0.6520106792449951), (u’Chris’, 0.6505492925643921), (u’Mark’, 0.6491551995277405), (u’Matt’, 0.6386727094650269), (u’Daniel’, 0.6294828057289124), (u’Greg’, 0.6267883777618408), (u’Jeff’, 0.6265031099319458)]

But there’s another way:  you can look at the words closest to “Jennifer” (which are essentially all first names) and pick out the ones which are closer to “he” than to “she”.  This gives

Matthew, Jeffrey, Jason, Jesse, Joshua.

>>> [x[0] for x in model.most_similar(‘Jennifer’,topn=2000) if model.similarity(x[0],’he’) > model.similarity(x[0],’she’)]
[u’Matthew’, u’Jeffrey’, u’Jason’, u’Jesse’, u’Joshua’, u’Evan’, u’Brian’, u’Cory’, u’Justin’, u’Shawn’, u’Darrin’, u’David’, u’Chris’, u’Kevin’, u’3/dh’, u’Christopher’, u’Corey’, u’Derek’, u’Alex’, u’Matt’, u’Jeremy’, u’Jeff’, u’Greg’, u’Timothy’, u’Eric’, u’Daniel’, u’Wyvonne’, u’Joel’, u’Chirstopher’, u’Mark’, u’Jonathon’]

Which is a better list of “male analogues of Jennifer?”  Matthew is certainly closer to Jennifer in word2vec distance:

>>> model.similarity(‘Jennifer’,’Matthew’)

0.61308109388608356

>>> model.similarity(‘Jennifer’,’David’)

0.56257556538528708

But, for whatever, reason, “David” is coded as much more strongly male than “Matthew” is; that is, it is closer to “he” – “she”.  (The same is true for “man” – “woman”.)  So “Matthew” doesn’t score high in the first list, which rates names by a combination of how male-context they are and how Jennifery they are.  A quick visit to NameVoyager shows that Matthew and Jennifer both peaked sharply in the 1970s; David, on the other hand, has a much longer range of popularity and was biggest in the 1950s.

Let’s do it again, for Susan.  The two methods give

David, Robert, Mark, Richard, John

Robert, Jeffrey, Richard, David, Kenneth

And for Edith:

Ernest, Edwin, Alfred, Arthur, Bert

Ernest, Harold, Alfred, Bert, Arthur

Pretty good agreement!  And you can see that, in each case, the selected names are “cultural matches” to the starting name.

Sidenote:  In a way it would be more natural to project wordspace down to the orthocomplement of “he” – “she” and find the nearest neighbor to “Susan” after that projection; that’s like, which word is closest to “Susan” if you ignore the contribution of the “he” – “she” direction.  This is the operation Ben Schmidt calls “vector rejection” in his excellent post about his word2vec model trained on student evaluations.  

If you do that, you get “Deborah.”  In other words, those two names are similar in so many contextual ways that they remain nearest neighbors even after we “remove the contribution of gender.”  A better way to say it is that the orthogonal projection doesn’t really remove the contribution of gender in toto.  It would be interesting to understand what kind of linear projections actually make it hard to distinguish male surnames from female ones.

Google News is a big enough database that this works on non-English names, too.  The male “Sylvie”, depending on which protocol you pick, is

Alain, Philippe, Serge, Andre, Jean-Francois

or

Jean-Francois, Francois, Stephane, Alain, Andre

The male “Kyoko” is

Kenji, Tomohiko, Nobuhiro, Kazuo, Hiroshi

or

Satoshi, Takayuki, Yosuke, Michio, Noboru

French and Japanese speakers are encouraged to weigh in about which list is better!

Update:  Even a little more messing around with “changing the gender of words” in a followup post.

Tagged , , , , , ,

Booklist 2013

This is not a typo — I was going to post about the books I read in 2015 but realized I’ve fallen out of the habit, and haven’t actually done a roundup since 2012! Here are the books of 2013:

 

  • 31 Dec 2013:  The Yacoubian Building, Alaa Al Aswany.
  • 17 Dec 2013: The Custom of the Country, Edith Wharton.
  • 29 Nov 2013:  Infinitesimal, Amir Alexander.
  • 19 Nov 2013:  The Simpsons and Their Mathematical Secrets, Simon Singh.
  • 2 Nov 2013:  The Panic Virus, Seth Mnookin.
  • 29 Oct 2013:  Taipei, Tao Lin.
  • 22 Oct 2013:  The Twelve, Justin Cronin.
  • 7 Oct 2013:  Fads and Fallacies in the Name of Science, Martin Gardner.
  • 15 Sep 2013:  The More You Ignore Me, Travis Nichols.
  • 11 Sep 2013:  Undiluted Hocus-Pocus:  The Autobiography of Martin Gardner.
  • 1 Sep 2013:  JoylandStephen King.
  • 27 Aug 2013:  The Ninjas, Jane Yeh.
  • 20 Aug 2013:  Time of the Great Freeze, Robert Silverberg.
  • 11 Aug 2013:  The Buddha in the Attic, Julie Otsuka.
  • 29 Jul 2013:  Lexicon, Max Barry.
  • 20 Jul 2013: Forty-One False Starts, Janet Malcolm.
  • 12 Jul 2013: Thinking in Numbers, Daniel Tammet.
  • 10 Jul 2013:  Boundaries, T.M. Wright.
  • 26 Jun 2013:  Let’s Talk About Love:  A Journey to the End of Taste, by Carl Wilson.
  • 15 Jun 2013:  Goslings, J.D. Beresford.
  • 1 Jun 2013:  You, Austin Grossman.
  • 25 May 2013:  The Night Land, William Hope Hodgson.
  • 10 May 2013:  20th Anniversary Report of the Harvard-Radcliffe Class of 1993
  • 5 May 2013:  The Vanishers, Heidi Julavits.
  • 17 Apr 2013:  Belmont, Stephen Burt.
  • 10 Apr 2013:  Among Others, Jo Walton.
  • 2 Apr 2013:  Math on Trial, by Leila Schneps and Coralie Colmez
  • 25 Mar 2013:  The Fun Parts, Sam Lipsyte.
  • 14 Mar 2013:  Mathematical Apocrypha, Steven Krantz.
  • 7 Mar 2013:  The Magic Circle, Jenny Davidson.
  • 2 Mar 2013: SnowAdam Roberts.
  • 24 Feb 2013:  A Hologram for the King, Dave Eggers.
  • 9 Feb 2013:  The Wind Through the Keyhole, Stephen King.
  • 8 Feb 2013:  The Life and Opinions of a College Class, the Harvard Class of 1926.
  • 15 Jan 2013:  When the Tripods Came, John Christopher.

 

34 books.  21 fiction, 11 non-fiction, 2 books of poetry (note to self:  at some point read a book of poems by a poet I don’t personally know.)  Of the novels, 8 were SF/fantasy.

Best of the year:  Impossible to choose between The Custom of the Country and Forty-One False Starts.  

Wharton often writes about the drive to acquire money and status, which she presents not as a means to meet other basic human needs (food, security, companionship) but as a basic need in itself, and pretty near the base of the pyramid.  Sometimes the particular situation is a little dated (as in the concern with divorce in Age of Innocence) but Custom of the Country, which is about a New York deformed by a sudden influx of new, uncivilized wealth absorbing everything around it, couldn’t be more topical.

Janet Malcolm is of course the best essayist alive.  Forty-One False Starts is a collection of pieces, mostly from the New Yorker I think, mostly new to me.  The title track is amazing:  just as it says, it’s 41 possible openings to an essay, each one abandoned as Malcolm tries to start again.  (Or maybe as Malcolm pretends to start again; was the collage her plan all along?  That would certainly make them “false starts” in the literal sense of the words.)  The same anecdotes appear in multiple sections, from multiple points of view, or rather, from the same point of view, Malcolm’s, which always seems to be viewing from everywhere at once.  Here’s the first paragraph from false start 3 (which is just two paragraphs long):

All during my encounter with the artist David Salle—he and I met for interviews in his studio, on White Street, over a period of two years—I was acutely conscious of his money. Even when I got to know him and like him, I couldn’t dispel the disapproving, lefty, puritanical feeling that would somehow be triggered each time we met, whether it was by the sight of the assistant sitting at a sort of hair-salon receptionist’s station outside the studio door; or by the expensive furniture of a fifties corporate style in the upstairs loft, where he lives; or by the mineral water he would bring out during our talks and pour into white paper cups, which promptly lost their takeout-counter humbleness and assumed the hauteur of the objects in the Design Collection of the Museum of Modern Art.

“assumed the hauteur”  I love.  The capitals of Design Collection and Museum of Modern Art I love.  And there’s the presence of money in New York and the anxiety it stirs into the world of for-lack-of-a-better-word “culture”, just as in Wharton.  And Wharton is in Forty-One False Starts, too, in Malcolm’s essay “The Woman Who Hated Women”.  In fact, I’m pretty sure it was that essay that spurred me to start reading Wharton again, which I’ve been doing on and off ever since.  Malcolm writes:

There are no bad men in Wharton’s fiction. There are weak men and there are foolish men and there are vulgar New Rich men, but no man ever deliberately causes harm to another person; that role is exclusively reserved for women.

As for The Custom of the Country:

With Undine Spragg, the antiheroine of ”The Custom of the Country” (1913), Wharton takes her cold dislike of women to a height of venomousness previously unknown in American letters, and probably never surpassed. Undine’s face is lovely, but her soul is as dingy as Gerty Farish’s flat. Ralph Marvell, one of her unfortunate husbands, reflects on “the bareness of the small half-lit place in which his wife’s spirit fluttered.”

I hate to disagree with Janet Malcolm.  But I disagree!  Back in 2013 I had a very well-worked out theory of this book, in which Undine Spragg was not particularly a villain, but rather the character who was best able to adapt to the new customs and the new country.  The men are weak, as Malcolm says, but indulgence of weakness can be a way of deliberately causing harm.  For every one of Undine’s “can’t believe she did/said that” moments in the book, there’s an analogous crime committed by one of the other characters, but expressed with more gentility.  Anyway, I’ve forgotten all my examples.  But it was a good theory, I promise!  I will admit that, having now read Ethan Frome, I can’t deny that there’s some extent to which Wharton experiences femaleness as a kind of horror.  But I don’t think that’s what’s going on with Undine Spragg.  (I also disagree with Roxane Gay about May Welland, who I totally think is meant by Wharton to be sympathizable-with but not likable compared with Countess Oleska, whose side I think Age of Innocence 100% takes, if it takes anyone’s.  Maybe more on this in the 2015 post.)

Others I should have blogged about:  I read Taipei because I was curious about Tao Lin, who some people think is a prankster masquerading as a fiction writer and other people think is really a fiction writer.  It’s the latter.  I mean, look at this map:

339382_10151905688102481_1749687214_o

He’s clearly somebody who sees himself in the tradition of experimental English-language fiction (Grace Paley!  Barthelme!  Stephen Dixon!  James freaking Purdy!) and I thought Taipei reflected that.  It was way more Barthelme than it was weird Twitter.  I had a good worked-out theory for this one, too, which I also forgot to blog.  Negative space:  it was a novel about a poet who is never seen writing or reading or performing poetry; i.e. a novel which places the experience of not-producing-poetry at the center of the poetic project.  Also there was something about the emphasis on Apple products and the relationship with China, where they’re produced — i.e. the novel is intently focused on use of Apple products while hiding the production of Apple projects, just as it’s intently focused on poetry while hiding the production of poetry.  But I was more into this interpretation before the novel actually goes to Taipei.  (And yes I know Taipei is not in the PRC; I felt willing to fudge the geography.)

Fads and Fallacies in the Name of Science:  from 1956, but, like Custom of the Country, almost painfully topical.  People don’t believe in orgone therapy anymore but the anti-scientific style in American culture is as healthy as ever.  Let’s Talk About Love:  the best book in existence about the problem of the “guilty pleasure,” or of art being “so bad it’s good,” or the basic difficulty of criticism of living culture:  is the critic’s job to tell you what to like and why to like it, or to understand why the people who like it like it?   (“Neither” is an OK answer here but let’s face it, these are the two leading candidates, unless “dispassionately analyze the class position of the work and the material circumstances of its production” still counts.)

How to succeed in business without really dying

The New York Times reports that people with a longer workweek have more strokes.

People who work 55 hours or more per week have a 33 percent greater risk of stroke and a 13 percent greater risk of coronary heart disease than those working standard hours, researchers reported on Wednesday in the Lancet.

The new analysis includes data on more than 600,000 individuals in Europe, the United States and Australia, and is the largest study thus far of the relationship between working hours and cardiovascular health.

If for some reason you’re looking to write a contrarian “opposition to universal healthcare from the left” editorial, start right here!  When health insurance is tied to employment, as in the US model, businesses have some incentive to avoid workplace environments that leave their employees broken husks likely to require expensive long-term late-life care.  Once you break that link, businesses are free to work people until they stroke out, with the cost externalized to the health care system.

(Of course, an actual left take on this would no doubt involve heavier regulation on businesses to mitigate unhealthy workplace practices, expanding on things like OSHA, child labor laws, etc., but let’s not let that get in the way of a contrarian spin!)

So… yeah

Lately CJ has a habit of ending every story he tells by saying

“So… yeah.”

I first noticed it this summer, so I think he picked it up from his camp counselors. What does it mean? I tend to read it as something like

“I have told my story — what conclusions can we draw from it? Who can say? It is what it is.”

Is that roughly right? Per the always useful Urban Dictionary the phrase is

“used when relating a past event and teller is unsure or too lazy to think of a good way to conclude it”

but I feel like it has more semantic content than that. Though I just asked CJ and he says it’s just his way of saying “That’s all.” Like “Over and out.”

So yeah.

Benson Farb’s ICM talk

One of the things I’ve been spending a lot of time on mathematically is problems around representation stability and “FI-modules,” joint with Tom Church, Benson Farb, and Rohit Nagpal.  Benson just talked about this stuff at the ICM, and here it is:

In the latest stable representation theory news, Andy Putman and (new Wisconsin assistant professor!) Steven Sam have just posted an exciting new preprint about the theory of representations of GL_n(F_p) as n goes to infinity; this is kind of like the linear group version of what FI-modules does for symmetric groups.  (Or, if you like, our thing is their thing over the field with one element….!)  This is something we had hoped to understand but got very confused about, so I’m looking forward to delving into what Andy and Steven did here — expect more blogging!  In particular, they prove the Artinian conjecture of Lionel Schwartz.  Like I said, more on this later.

Hey, what’s that book you’re not reading?

bookfreshpressIn the Wall Street Journal this weekend I define a new metric aimed at identifying books people are buying but not reading.

How can we find today’s greatest non-reads? Amazon’s “Popular Highlights” feature provides one quick and dirty measure. Every book’s Kindle page lists the five passages most highlighted by readers. If every reader is getting to the end, those highlights could be scattered throughout the length of the book. If nobody has made it past the introduction, the popular highlights will be clustered at the beginning.

Thus, the Hawking Index (HI): Take the page numbers of a book’s five top highlights, average them, and divide by the number of pages in the whole book. The higher the number, the more of the book we’re guessing most people are likely to have read. (Disclaimer: This is not remotely scientific and is for entertainment purposes only!)

At the end I suggest we call this number the Piketty Index instead, because Piketty’s unlikely megahit Capital in the Twenty-First Century comes in with an index of 2.4%, the lowest in my sample.

But it’s not the winner anymore!  Hillary Clinton’s Hard Choices scores an amazing 1.9%.  But somehow I feel like HRC’s book is in a different category entirely; unlike Piketty, I’m not sure I believe it’s a book people even pretend to intend to read.

The piece got lots of press, including a nice writeup at Gizmodo today.  So I thought I’d add a few more comments here, to go past what I could do in an 800-word story.

  • Lots of people asked:  what about Infinite Jest?  In fact, that book was in the original piece but got cut for length.  Here’s the paragraph:

    Infinite Jest, by David Foster Wallace.  HI 6.4%.  There was a time, children, when you couldn’t ride the 1/9 without seeing a dozen recent graduates straining under the weight of Wallace’s big shambling masterpiece.  Apparently it was too heavy for most.  Yes, I included the endnotes in the page count.  This is another one whose most famous line – “I am in here” – doesn’t crack the Kindle top five.

  • Other books I computed that didn’t make it into the WSJ:  Stephen King’s new novel Mr. Mercedes scores 22.5%.  How To Win Friends and Influence People gets 8.8%.  And How Not To Be Wrong comes in at 7.7%.  In fact, the original idea for the piece came from my dismay that all the popular highlights in my book were from the first three chapters.  But actually that puts How Not To Be Wrong in the middle of the nonfiction pack!
  • Important:  I highly doubt the Piketty Index of the book is actually a good estimate for the median proportion completed.  And I think different categories of books can’t be directly compared.  All nonfiction books scored lower than all novels (except Infinite Jest!)  I don’t think that’s because nobody finishes nonfiction; I think it’s because nonfiction books usually have introductions, which contain lots of direct assertions and thesis statements, exactly the kind of thing Kindle readers seem to like highlighting.
  • Challenges:  can you find a book other than The Goldfinch whose index is greater than 50%?  Can you find a nonfiction book which beats 20%?  Can you find a book of any kind that scores lower than Hillary Clinton’s Hard Choices?

 

Sympathy for Scott Walker

The Milwaukee Journal-Sentinel suggests that the slow pace of job creation in Wisconsin, not recall campaign shenanigans, may be Scott Walker’s real enemy in his upcoming re-election campaign:

In each of Walker’s first three years, Wisconsin has added private-sector jobs more slowly than the nation as whole, and the gap is sizable. Wisconsin has averaged 1.3% in annual private-sector job growth since 2010; the national average has been 2.1%. Wisconsin’s ranking in private-sector job growth was 35 among the 50 states in 2011, 36 in 2012 and 37 in 2013.

Combining the first three years of Walker’s term, the state ranks behind all its closest and most comparable Midwest neighbors: Michigan (6 of 50), Indiana (15), Minnesota (20), Ohio (25), Iowa (28) and Illinois (33).

I think this is slightly unfair to Walker!  Part of the reason Michigan is doing so well in job growth since 2010 is that Michigan was hammered so very, very hard by the recession.  It had more room to grow.  Indiana’s unemployment rate was roughly similar to Wisconsin’s in the years leading up to the crash, but shot up to 10.8% as the economy bottomed out (WI never went over 9.2%.)  Now Indiana and Wisconsin are about even again.

But I do mean slightly unfair.  After all, Walker ran on a change platform, arguing that Jim Doyle’s administration had tanked the state’s economy.  In fact, Wisconsin weathered the recession much better than a lot of our neighbor states did.  (The last years Wisconsin was above the median for private-sector job growth?  2008 and 2010, both under Doyle.)   There’s some karmic fairness at play, should that fact come back to make Walker look like a weak job creator compared to his fellow governors.

 

 

Tagged , ,

Charles Franklin on “Data Visualization in Political Science,” Wednesday at Union South

Good talk:  my colleague Charles Franklin, currently on leave at Marquette running serious polling on Wisconsin’s weird political microclimate, is back in town this Wednesday to give a talk about data visualization in political science:

Polling Your Resources: Using Data Visualization in Political Science

  • DateWednesday, April 11, 2012
  • Time3 p.m.
  • LocationTITU, Union South
  • DescriptionUW-Madison faculty member Charles Franklin will share examples of data visualization and discuss helping students and the public make sense of political data. If you plan to use data visualization in your teaching, come and learn how Franklin has honed this topic for his course Understanding Political Numbers. Franklin’s academic research focuses on advanced statistical and graphical analysis of public opinion and election outcomes. Light refreshments will be served. Presented by Engage.

Bovine fraternal skin graft

Another thing I learned from the August 1951 issue of The Times Review of the Progress of Science is that cows can accept skin grafts from their fraternal twins, but humans can’t.  That’s because cow fetuses actually share some blood and tissue in the womb, and automatically get desensitized to those particular foreign entities when they’re young enough not to reject them.  This was totally new to me but apparently if I knew anything about immunology I would already be familiar with this, because Peter Medawar’s work on the phenomenon earned him the 1960 Nobel Prize and more or less launched the field of acquired transplantation tolerance.

There was also an anecdote about a baby switched at birth, who doctor proved to be the identical twin of another child in his birth family by grafting a patch of his skin onto the other kid!

Tagged , ,
Follow

Get every new post delivered to your Inbox.

Join 677 other followers

%d bloggers like this: