Douglas Hofstaedter, ambigrams and gridfonts

Posted on April 25, 2006
Filed Under Uncategorized | Comments

I was reminded of Douglas when I saw the BoingBoing ambigram. Douglas introduced us to the practice of ambigrams (credit to Scott Kim) when I took courses with him as an undergraduate at the University of Michigan. (Here’s something really quite mindblowing – we’re talking 18 years ago!) Anyway, these were very interesting courses offered through Michigan’s great cognitive science department. Douglas was at the peak of his fame, having just published his magnum opus, Godel Escher Bach: An Eternal Golden Braid. I think he was on sabbatical from Indiana University at the time.

gridfont grid

One of the exercises in that class made an incredible impact on me. We were asked to design fonts… but with the constraint that the fonts had to be executed on a 2×3 grid, connecting only adjacent dots. And no cheating – you couldn’t use any embellishments (for instance the thickness of a line, or color of a line, go for a “long diagonal”, or anything of that sort..) A few other rules were imposed – we were to design just the lower case version (no capitals or punctuation marks, etc.) The lower case letter “a” was to be the typeface version not the handwritten version, (i.e. like this “a”, not like this “a”.) There were probably a few other rules that have been lost in the passage of time.

The first time you try to design a font, you run straight up against the absurd constraint of the grid. It’s an absurdly small footprint that leaves very little room for “creativity.” Just executing the alphabet against this backdrop is an accomplishment.

Later, as you execute your fifth and tenth and twentieth font from start to finish, you begin to attain some level of craftsmanship. You begin to discover the relationship between letters, i.e. doing the “b” this way is going to have obvious implications for the “d”. You begin to describe the fonts in various ways, and creativity and style begins to rear its head.

Font 1
Font 1


For example, Font 1 (and I’ve just executed a-c and s-u as examples) is a highly stylized serifed font with lots “diagonalness.” The “diagonal” theme is evident throughout. A subtheme might be the disconnectedness of the “s”. Perhaps I could have resonated that theme against the “a”, omitting the segment that connects dot 6 to dot 9. As “font designer” (he said puffing himself up,) I chose not to, but concede it would have been a reasonable option.

Font 2
Font 2
Font 3
Font 3

Fonts 2 and 3 also has an obvious themes, and interestingly the “t” and “u” designs overlap between them. I’m dubious about the “u”’s and perhaps if I followed through and designed the “v”’s it would have led to significant changes.

In Douglas’s class, we’d sit there and review each other’s gridfonts for hours. We’d question design choices, labor over the tiniest of lines and the “grave” implications it would have for other letters in the alphabet.

The gridfont exercise bears many gifts. Working in a world of absurd reductionism, the essence of design, style, and creativity emerge in zen-like moments of insight. It’s as if other approaches toward design philosophy were “Newtonian,” and gridfonts was an electron microscope that revealed the quantum building blocks of creativity. This post probably won’t make a lot of sense unless you get out the graph paper and invest the energy to actually follow through on the exercise. Recommended!

Ambigrams are another lovely way of introducing these kinds of constraints. There are many flavors of ambigram, that exhibit any variety of symmetry. Check out Scott Kim’s page for more. Again, ambigrams are kinda fun and novel to look at but any real benefit is derived from trying to construct them. It’s another great way to exercise muscles you’ve forgotten you have…

They say, “Necessity is the mother of invention”. Now that I’m deliberately contemplating that old saw, I’m reminded that it’s a multi-faceted statement. I’ve always taken the primary sense to be solutions follow need… As Y Combinator’s motto says, “Make something people want!” (That’s the greatest motto an incubator could have IMHO.) But there’s another sense… Fat and happy doesn’t breed creativity. Constraints breed creativity. Nobody builds a catapult out of bubble-gum and baling wire if they don’t have to… Now go listen to the “Mothers of Invention” and find out what creativity really is…

Comments

5 Responses to “Douglas Hofstaedter, ambigrams and gridfonts”

  1. Alex Jaffe on April 25th, 2006 8:21 pm

    I just want to say: You took classes with Douglas Hofstaedter? Wow. It is incredibly hard to find information on his career outside of the books online, but I did know he taught… Also, I’ve never seen the font exercise anywhere before, (that I know of) but I practice it EXACTLY as you described as a sort of alternative to doodling. That must be pretty primal to, er, written-language-using humans.

  2. Jon Udell on April 26th, 2006 1:04 pm

    Wow, indeed. It’s too bad I was at the Univ. of Mich at the wrong time, and in the wrong department.

    My favorite Hofstadter book is actually the lesser known Fluid Concepts and Creative Analogies. And here’s a nice connection. Gary McGraw, who is CTO of Cigital, did the LetterSpirit project which is chap 10 of that book. I asked him about it, and about his experiences during the Hofstadter cogsci heydey, in a recent podcast:

    http://weblog.infoworld.com/udell/2006/04/07.html

    LetterSpirit was a program that would be shown a letter in a gridfont, deduce what letter it was, deduce the style of the font, and then create the rest of the font.

  3. Bradley Horowitz on April 26th, 2006 3:24 pm

    Yes, I remember this program being discussed. Likely all the work in hammering out gridfonts by hand was really a ruse… makework foisted upon lowly students to input ground truth into the system! But had a huge impact on me nonetheless.

    A program that was shown “a” letter would have a very hard time deducing the font methinks. Totally underconstrained, especially if the letter is for instance an “i”. I’d imagine you’d need a critical mass of letters. You could probably get away with three if you selected the right three.

    I vaguely recall hearing about a “carny” who touted “I’ll tell you where you’re from within 500 miles just from hearing you pronounce the word ‘water’.” Water was a clue word that was heavily laden with signal (i.e. highly differentiated between accents.) Some letters are more telling than others.

  4. Rajat Paharia on April 26th, 2006 8:40 pm

    Scott’s work and the ambigrams in GEB and Dan Brown’s Angels and Demons inspired me to create an ambigram for my wife and I when we got married. It was a fun exercise. Check it out (Rajat + Laura): http://www.rootburn.com/wedding/rootburn.swf

    I remember attending a great panel on AI & Humor at Stanford years ago with Doug, Steve Martin, Marvin Minksy and Kim Binstead (University of Edinburgh, had written a program that came up with jokes).

  5. Chris Nicholls on April 7th, 2007 5:24 pm

    I also went to the Universtiy of Mich. around the same time as you Bradley, and I remember hearing (and taking part in) many interesting and widley discussed conversations about Douglas Hofstaedter, and his methods of conveying writing philosophy to todays English pronuciation. It’s funny to think that back then I had little interest in the matter, yet today I am a succusful Marine Biologist. I remember Hofstaedter saying that, “In time, underwater sea creatures will begin to develope seaking patterns similare to humans.” The fact that sea creatures can’t allready speak English amazes me… I’d love to hear you’re thoughts on the matter.

Leave a Reply