Gibson's "Virtual Light"

It’s hard not to be a fan of Gibson, if you’re working among the technorati. Our “sociologist of the near future” has predicted and coined a number of cultural/technological phenomena, and in “Idoru”, it seems to be the rise of big data. I’m reading it now, having just finished his “Virtual Light” a few weeks ago. The following passage strikes me not just because of its prescience, but its relevance to my own neurochemical disposition – a disposition that others in the field must certainly share.

 

 

 

 

 

“Coffee,” Laney said.

Laney was not, he was careful to point out, a voyeur. He had a peculiar knack with data-collection architectures, and a medically documented concentration-deficit that he could toggle, under certain conditions, into a state of pathological hyperfocus. This made him, he continued over lattes in a Roppongi branch of Amos ‘n’ Andes, an extremely good researcher. (He made no mention of the Federal Orphanage in Gainesville, nor of any attempts that might have been made to cure his concentration-deficit. The 5-SB trials or any of that.)

The relevant data, in terms of his current employability, was that he was an intuitive fisher of patterns of information: of the sort of signature a particular individual inadvertently created in the net as he or she went about the mundane yet endlessly multiplex business of life in a digital society. Laney’s concentration-deficit, too slight to register on some scales, made him a natural channel-zapper, shifting from program to program, from database to database, from platform to platform, in a way to was, well, intuitive.

And that was the catch, really, when it came to finding employment: Laney was the equivalent of a dowser, a cybernetic water-witch. He couldn’t explain how he did what he did. He just didn’t know.

I think that the correlation between developers and caffeine (and beer, and other substances) speaks to a real need for a “pathological hyperfocus” when doing computer research. Dive into any CS textbook or programming forum worth its salt, and you’ll see it right away: free software is “free as in beer”, according to Richard Stallman. The Java language is named for coffee, “said to be consumed in large quantities by the language’s creators.” Commentors on StackOverflow or Slashdot often jokingly offer each other beer in return for digital services. Substances are not just popular among developers, they are part of us, and part of the mythology of technology that we surround ourselves with.

At work, I drink several cups of coffee before any long project. Like liquid Adderal, coffee gets me in that pathological hyperfocus that Gibson writes about. I think it’s safe to say that coffee doesn’t just help me work, but I need it to work effectively. 

At the same time, people like Laney, myself, and others with ADHD seem to be predisposed to hyperfocusing naturally. Additionally, people with attention deficit disorders are at a higher risk for substance addiction. This begs the question – does the association between beer and caffeine indicate higher rates of attention deficit disorders  (because of chemical addictions) among developers?

One of my best photographs is featured on the main header.

NLP has a lot of potentional to transform the pharmaceutical industry

News today from a few sources that Cambridge, UK based Linguamatics has sealed a deal with Selventa, “a personalized healthcare company focused on stratification of patients and development of predictive biomarker panels based on disease-driving mechanisms”.  The official word is that Linguamatics is going to use their NLP engine to mine healthcare literature to “efficiently extract complex life science knowledge in a computable, structured, biological expression language (BEL)… that can be used to interpret large-scale experimental data”.

I was fortunate to meet a brilliant researcher from Selventa this September at a Cambridge Semantic Web Meetup at MIT. Dr Julian Ray gave a talk on the biological expression language (BEL). As the two companies have only just inked this deal, it was probably before they were licensing any Linguamatics lingware, but I was definitely impressed by what I saw. Selventa already brings some impressive technology to the table.

Reading between the lines, I think I have an idea of what’s going on here. Selventa’s model is to plug into a huge database of pharmaceutical research, and then search for statistical correlations between symptoms (sore throat, etc), existing pharmaceutical solutions (Tylenol, aspirin, etc), and proteins that occur often with these in the literature. The idea is that you can actually predict which new drugs will be effective treatments for symptoms simply by looking for compounds that occur often with them in experimental data, without actually doing any testing. Dr Ray called this predictive ability “the holy grail” of bioinformatic techology, because pharmaceutical companies could save shitloads of money and preepmt market trends. Unfortunately, this description of the technology underlines Selventa’s difficulties with performing good NLP on their sources. His presentation attracted a lot of criticism because of the low precision it got against the target corpus, and the possibilties for predicting the wrong (and potentially deadly) treatment for a symptom.

My guess is that they’re contracting Linguamatics to add the the British company’s secret sauce to their existing platform, so that they can achieve better precision and hopefully reach this holy grail of pharmaceutical text analysis. More power to them, I say – but they should watch out for other Cambridge startups like Entagen, who are already demonstrating more impressive results. More on them later.

[HiveFire NLP] via [Semantic Web] via [Drug Discovery and Development]

Those kids were really talented.

The American team holding court in first place.

I’m inspired. I can’t tell you how many teenage polyglots I met at CMU who spoke, variously, Icelandic, or Nahuatl, or Basque, or Kutchi, or barcode. An eighteen year old Russo-Canadian who spoke nine languages fluently; an eighteen year old Israeli-American who spoke Inuit, Basque, and dreamed in code; an eighteen year old from Pittsburgh who graduated high school at sixteen and is now interning at CMU, writing a pan-Bantu morphological analyzer. It’s hard not to be intimidated by prodigies like these, the smartest high schoolers in the world, and yet, here, now, I’m just inspired.

There was a long time in high school when, in retrospect, I was debilitatingly awkward and depressed and poured all of my angst into languages. I used to desperately live linguistics. Recently, over the last couple of years, I’ve hit a groove, and I think I got lazy, and I stopped being hungry. The teenage linguists at CMU reconnected me with that curiosity, and I’m ready to learn again.

Chillin' with the Vietnamese.

My clean installation of OS X Lion is running well so far, but I do have a problem with the mobile/grandma iOS style they’re bringing to the desktop. Xcode installs from the App Store now, nowhere else. I had to restart my installation of that ‘app’ halfway through, and needed to trash the aborted files before I started it up again. Lion would rather I simply ‘tapped’ and ‘clicked’ the half-dead ‘app’ away into the ‘cloud’ from Launchpad. Now I’m stuck with Xcode’s tell-tale heart, permanently buried in the floorboards of my /bin, occasionally reminding me that this system, although convenient, is inexpertly implemented.