What would a bot buy from Amazon, if given life—and a gift card loaded with credit? Noam Chomsky's Cartesian Linguistics, apparently.

It's hard to believe that'd be a random choice, but it is, coming from a creature engineered for randomness by a man fascinated with randomness -- and consumerism. My friend Darius Kazemi, Boston-based developer extraordinaire, has a long-held interest in randomness. He's made the Twitter account @metaphorminute, designed to tweet a random metaphor every couple minutes, and OutSlide, which generates a random set of slides based on phrase-oriented Google image results.

With a background primarily in games, he's always been drawn to roguelikes and other games where random generation is a factor in the experience; he's attracted to the idea of "abdicating design decisions to a computer."

For example, he recently noticed an apparently-random area of Manhattan where real estate seems particularly expensive; for some reason, trading computers have superior latency there, leading financial firms to buy up real estate all to gain space for a couple of extra machines and the efficiency thereof. The whim of a machine caused an unpredictable spike in the value of a certain spot on the landscape.

"I like randomness because it's telling you straight up that there's a computer making this decision, and it's completely alien," Kazemi tells me over the phone. "It's based off no criteria that you'd ever use in your own life."

In the recent year he and his spouse have bought a house, and with it comes increased thought on the conscientious couple's part to ideas about consumerism, "things." Kazemi noticed how the occasional sudden arrival of back-ordered Amazon products he'd long since forgotten about ordering feels somehow more exciting, "like a gift you bought yourself," and wondered what it would feel like to design a program that buys you things seemingly at random?

The bot's purpose, in Kazemi's words, is largely to "fill [his] life with crap," to see if somehow those purchases feel more or less meaningful than something he would have conscientiously chosen himself; a way, if you will, of exploring his attachment to that "crap."

Thus Random Shopper was born, complete with controls that keep it from buying anything too expensive or too physically large (spouse Courtney was "supportive," Kazemi says, but "was also like, 'I don't want skis showing up at the house.'"). Random Shopper has its own Amazon account, and its budget is limited to a set amount on a gift card. For now, Kazemi's restricted its categories to CDs, DVDs and paperback books -- that keeps the size issue under control, and limits purchases to stuff that's easily digitized, consumable and can be given away or donated, "as opposed to, like, a plug for a device that I don't own," he explains.

The bot shops using a random word plucked from the Wordnik API. Since Kazemi is able to run simulations on the bot up to the point of actual purchase, he plans to experiment with other categories, like housewares, just to see what kinds of things the bot would send.

"It's like having a martian as a personal shopper," he reflects.

The first time he turned on the bot, Kazemi eagerly awaited his first shipment, which he knew would come the following week. "When I saw the Chomsky book, I laughed," he says. "My AI just sent me a Chomsky book; that's hilarious, because Chomsky did a lot of work that was instrumental in the early formulation of AI."

The second gift: A fingerprint-logo black CD case that simply read "Ákos Rózmann" on the front, a name Kazemi had never heard. "The first track is this really abrasive noise, and it sounds like there's something wrong with the CD, and I was going, 'well, either this is some very avant-garde music, or I got a defective CD...' by the second track, I actually really liked it, and I was smiling ear to ear," he enthuses. "It was like, 'my bot sent me an awesome present.'"

Personifying bots is easy and comes naturally, Kazemi reflects. When it comes to his @metaphorminute bot, "I certainly consider it to be like a child. Not a child I'm particularly attached to, and if it died, I wouldn't cry. But maybe to the extent your pet goldfish is like a child, and you are responsible for it," he says. Once, @metaphorminute accidentally used foul language, tasking the parent with teaching it how to talk politely. "I do sort of casually refer to them like you'd refer to a child -- 'one of my bots did the cutest thing today!'"

Since that initial purchase, Random Shopper has sent along The Oxford History of World Cinema, 1995 sci-fi film Screamers, and something called the Covenant Discipleship Parents' Handbook. And has drawn some criticism, too, of the developer's leisure to spend real money on a bot that always runs the very real risk of essentially wasting money.In a blog post Kazemi says he recognizes the validity of that criticism, but likens it to the cost artists invest in supplies or research.

And there is an element of very conscious subversion to Random Shopper: "I like the idea of jamming Amazon's recommendations slightly, by having a consumer that doesn't conform to any statistical models," he says. "It's a tiny subversion, but I like that idea. I have a friend who was obsessed with putting nonsensical information in his Facebook profile just to throw off their predictive algorithms a bit. It's kind of like that."

Theoretically, if a mass of people changed all their Facebook data to nonsense, or set random shopper armies loose on Amazon, it'd break these services' growing ability to know us through data, to target and market to people with increasingly-unnerving, ever more personalized aptitude.

"Something else that's interesting to me is that within randomness, there's the idea of apophenia -- the human tendency to find patterns where there are none," Kazemi notes, pointing to how people once saw gods in the patterns of stars, or see deities in stains and mottles. "It'll be interesting to see how my relationship to this stuff evolves."

"Mostly it comes down to my weird sense of humor," he adds. "It's not that technically challenging to do this stuff."

I can haz transformational grammar?

Via Justin Bernacki and Trust me, I'm a linguist.

Languages come and go and blend. It's likely been that way forever and the process only accelerates under the influence of mega-languages (like English) that represent a sort of global means of communication. But, increasingly, people who are at risk of losing their native language entirely are fighting back—trying to encourage more people to be bilingual and save the native language from extinction.

At Discover Magazine, Veronique Greenwood has a really interesting story about a mathematician who is helping to preserve Scottish Gaelic. How? The researcher, Anne Kandler, has put together some equations that can help native language supporters target their programs and plan their goals.

Some of the numbers are obvious—you must know how many people in the population you’re working with speak just Gaelic, how many speak just English, and how many are bilingual, as well as the rate of loss of Gaelic speakers. But also in the model are numbers that stand for the prestige of each language—the cultural value people place on speaking it—and numbers that describe a language’s economic value.

Put them all together into a system of equations that describe the growth of the three different groups—English speakers, Gaelic speakers, and bilinguals—and you can calculate what inputs are required for a stable bilingual population to emerge. In 2010, Kandler found that using the most current numbers, a total of 860 English speakers will have to learn Gaelic each year for the number of speakers to stay the same. To her, this sounded like a lot, but the national Gaelic Development Agency was pleased: it’s about the number of bilingual speakers they were already aiming to produce through classes and programs.

Read the rest at Discover Magazine

The results are interesting. The phrases themselves turn out to be significantly distinctive, meaning they're made up of combinations of words that are unlikely to appear in the corpus. By contrast, memorable phrases tend to use very ordinary grammatical structures that are highly likely to turn up in the corpus.

They also found that memorable phrases tend to use pronouns (other than you), the indefinite article a rather than the definite article the, and verbs in the past rather than present tense. These are all features that tend to make phrases general rather than specific.

So memorable phrases contain generic pearls of wisdom expressed with unusual combinations of words in ordinary sentences.

The Secret Science of Memorable Quotes ]]> http://boingboing.net/2012/04/06/why-certain-phrases-are-memora.html/feed 13 http://boingboing.net/2012/03/13/passphrases-suck-less-than-pas.html http://boingboing.net/2012/03/13/passphrases-suck-less-than-pas.html#comments Tue, 13 Mar 2012 23:24:06 +0000 Cory Doctorow http://boingboing.net/?p=148959 In "Linguistic properties of multi-word passphrases" (PDF, generates an SSL error) Cambridge's Joseph Bonneau and Ekaterina Shutova demonstrate that multi-word passphrases are more secure (have more entropy) than average user passwords composed of "random" characters, but that neither is very secure. In a blog post, Joseph Bonneau sums up the paper and the research that went into it.

Some clear trends emerged—people strongly prefer phrases which are either a single modified noun (“operation room”) or a single modified verb (“send immediately”). These phrases are perhaps easier to remember than phrases which include a verb and a noun and are therefore closer to a complete sentence. Within these categories, users don’t stray too far from choosing two-word phrases the way they’re actually produced in natural language. That is, phrases like “young man” which come up often in speech are proportionately more likely to be chosen than rare phrases like “young table.”

This led us to ask, if in the worst case users chose multi-word passphrases with a distribution identical to English speech, how secure would this be? Using the large Google n-gram corpus we can answer this question for phrases of up to 5 words. The results are discouraging: by our metrics, even 5-word phrases would be highly insecure against offline attacks, with fewer than 30 bits of work compromising over half of users. The returns appear to rapidly diminish as more words are required. This has potentially serious implications for applications like PGP private keys, which are often encrypted using a passphrase. Users are clearly more random in “passphrase English” than in actual English, but unless it’s dramatically more random the underlying natural language simply isn’t random enough. Exploring this gap is an interesting avenue for future collaboration between computer security researchers and linguists. For now we can only be comfortable that randomly-generated passphrases (using tools like Diceware) will resist offline brute force.

Some evidence on multi-word passphrases (via Schneier) ]]> http://boingboing.net/2012/03/13/passphrases-suck-less-than-pas.html/feed 33 http://boingboing.net/2012/01/05/every-writing-system-ever-pr.html http://boingboing.net/2012/01/05/every-writing-system-ever-pr.html#comments Thu, 05 Jan 2012 14:00:16 +0000 Cory Doctorow http://boingboing.net/?p=137379
Omniglot is an intimidatingly complete site devoted to cataloging every writing system that ever existed. As JoshP says, "If you ever need to transliterate Punic... this is the place."

Omniglot - the guide to languages, alphabets and other writing systems ]]> http://boingboing.net/2012/01/05/every-writing-system-ever-pr.html/feed 12 http://boingboing.net/2011/12/28/linguistics-turing-completene.html http://boingboing.net/2011/12/28/linguistics-turing-completene.html#comments Thu, 29 Dec 2011 07:35:10 +0000 Cory Doctorow http://boingboing.net/?p=136369

Yesterday's keynote at the 28th Chaos Computer Congress (28C3) by Meredith Patterson on "The Science of Insecurity" was a tour-de-force explanation of the formal linguistics and computer science that explain why software becomes insecure, and an explanation of how security can be dramatically increased. What's more, Patterson's slides were outstanding Rageface-meets-Occupy memeshopping. Both the video and the slides are online already.

Hard-to-parse protocols require complex parsers. Complex, buggy parsers become weird machines for exploits to run on. Help stop weird machines today: Make your protocol context-free or regular!

Protocols and file formats that are Turing-complete input languages are the worst offenders, because for them, recognizing valid or expected inputs is UNDECIDABLE: no amount of programming or testing will get it right.

A Turing-complete input language destroys security for generations of users. Avoid Turing-complete input languages!

Patterson's co-authors on the paper were her late husband, Len Sassaman (eulogized here) and Sergey Bratus.

LANGSEC explained in a few slogans ]]> http://boingboing.net/2011/12/28/linguistics-turing-completene.html/feed 49 http://boingboing.net/2011/12/09/the-academic-linguistics-of-lo.html http://boingboing.net/2011/12/09/the-academic-linguistics-of-lo.html#comments Fri, 09 Dec 2011 17:34:53 +0000 Maggie Koerth-Baker http://boingboing.net/?p=133633

On Submitterator, Musicman pointed me towards this great presentation on LOLspeak as a form of language play, and why people engage in that play. According to Lauren Gawne, who gave this speech last week at the Australian Linguistics Society conference, the choice to use LOLspeak has a lot to do with establishing identity—the playful identity of "cat", and the serious identity of "knowledgeable Internet user".

Includes an explanation of why LOLspeak is language play and not some language mashup "kitty pidgin".

You can read more about this on Lauren Gawne's blog Superlinguo.

The video, by the way, is 20 minutes long. It's also got a little bit of weird, warbly feedback in the audio, but that doesn't get in the way of hearing what Gawne is saying.

Last year, I stumbled across some of the cool history of American Sign Language, documenting how it evolved out of both formal and informal languages—systems Deaf children used to communicate at home, and the systems they were taught as Deaf schools drew diverse groups from a wide geographical range. For American Sign Language, this process happened in the 19th century. In other parts of the world, it's still ongoing. For instance, in Nicaragua, Deaf people who are in school now are learning a much more formalized language, with a much bigger vocabulary, than those who went to school in the 1980s.

Those international differences are fascinating to me, so I'm really pleased to find this post on the Sinosplice blog, discussing the Chinese system of finger spelling. The blogger there is a linguist, so there's a lot of neat perspective in the linked post and others on the linguistic mechanics of finger spelling and sign language in China.

Finger spelling is very different from a sign language. In a sign language, you'd have one hand movement or hand position that stands for the concept "bird." In finger spelling, you'd have several different movements/positions for each letter or sound of the word "bird." You probably picked up some American finger spelling from Sesame Street, it's likely to at least look somewhat familiar. But the really cool thing about this post, is that it contrasts that system with the finger spelling alphabets used in Russia, Japan, and several that have been used historically in China. That's the US system above. Below, the modern Chinese system that corresponds to the pinyin, a way of transcribing printed Chinese words into Roman letters.

Other features include ‘Canadian raising’, the most-widely known Canadian pronunciation feature. Canadian raising affects the diphthongs in words such as wife, price or life and house, about or shout. Canadian pronunciations, though far from universal, are often perceived as weef instead of wife and a boot instead of about by outsiders. There are also other, less well-known Canadian differences, such as the Canadian integration pattern of foreign sounds represented by<a>. In words like pasta, lava, plaza, and drama the foreign <a> sound acquires the vowel in father in American English and British English, but the vowel of cat in Canadian English. Canadian English

(Image: Canada, a Creative Commons Attribution (2.0) image from alexindigo's photostream) ]]> http://boingboing.net/2011/08/08/howto-sound-canadian.html/feed 51