To follow this blog by email, give your address here...

Saturday, April 09, 2005

Focus, Garry Kasparov, Isaac Newton, AGI, Business Management, Episodic Memory, Buddhists Who Don’t Spill Tea in Hats, etc.... and, oh yeah, Focus....

Well, this is an odd blog entry to post, because I wrote it (except this new introductory babbling) a couple days ago and it doesn't really reflect my mood or thoughts at this moment very well. Right now I'm in a quite perky mood, just about to go outside and play some tennis on a sunny Saturday, and then come back in afterwards and launch into the final revisions on "The Path to Posthumanity" (a book on the future that I wrote a couple years ago, and now need to finish in a hurry since the publisher decided to light a fire under my butt by listing the book on amazon.com and getting some sales of the as-yet not-quite-existent book...). Hopefully I can finish these revisions in the next week or so (in spare time, since most of my time is spent on software-biz stuff these days) and then plunge into the almost-done Novamente book that I really wish I were able to find more time for...

But anyway, here is the long blog entry that I wrote at the start of a business trip a few days ago, but didn't find time to post until now. A lot of rambling nonsense I'm afraid, but also some interesting nuggets here and there.

By the way, a couple people have emailed me to ask about Captain Zebulon's famous turtle tank. I not only changed it but I replaced the filter with a new, much better Penguin-brand filter that seems much more effective at filtering out the massive amounts of crap that water-turtles produce as compared to fish (thus hopefully reducing the need for frequent manual tank-cleaning). However, the new filter makes quite a loud noise -- you can hear it in the background of my son's latest musical recordings, of his soon-to-be-classic tune "The King of the Jews is Singing the Blues." (Unfortunately, his recording only exists within a videogame he's creating using RPGMaker 2003, so I can't post it here. Trust me, though, it's good. If I could sing as well as Zeb, I'd give up AI and become the next Michael Jackson. Er ... well ... something like that....)

OK OK, here is the biz-trip-blog...

I’m writing these words on my laptop at Gate B44 in the Washington Dulles airport – I missed my 6:10 AM flight due to stupidly forgetting to reset my alarm clock back to Standard Time from Daylight Savings time (so it woke me up at 5AM rather than 4AM). Rescheduled for a 7:40AM flight, I’ve got a bit of extra time in the airport -- so I took 15 minutes and speed-read a business/management book in the airport bookstore. (It’s not a vacation flight, unfortunately – I had a great vacation w/ the wife and kids last week, swimming and snorkeling and canoeing in South Florida, but this is a one-day business trip to California, to meet with some folks potentially interested in funding Biomind … (my bioinformatics business – which, after a few years of work, might eventually yield me enough profit that I can pay myself and a small team enough money to build the fabled Thinking Machine whose design lays mostly-neglected on my desktop….)). Not my usual fare, though I’ve probably read a few dozen business/management books in my life, but this one was moderately interesting. (As always with such books, the core information could be summarized in about 5 pages, but there are lots of evocative anecdotes. This ties in with something I’ve often thought about in the context of the Novamente design: human episodic memory seems to be at least partially organized by “story.” The human brain seems to store episodes different from procedures or declarative knowledge, and it seems to store them in units defined by some sort of conceptual coherence. In Novamente a “story” corresponds to a particular kind of “map,” meaning a set of nodes and links that are joined by HebbianLinks mutually reinforcing each other; a story differs from a generic map in that the nodes and links within it pertain to a set of events unfolding over time.)

But anyway … I digress (which is the main amusing thing about blogging – unlike in the “serious” writing I do, I allow myself to ramble and digress almost unlimitedly. I used to do that in writing fiction, but in the novel I’m writing now I’m orchestrating the digressions in a more careful way, which results in a better product but a less relaxed writing process. True, Jack Kerouac and Philip K. Dick wrote a lot of great stuff via pure “downhill skiing”, and Kerouac allowed a lot of digression in his writing-skiing process, but I don’t seem able to control my writing-skiing as well as those guys in real-time – my real-time verbal-conceptual improvisation is too wide-ranging and whacky, and it needs rational-critical post-processing to be made into something really artistic … unless the (err…) “artistic” nature sought is that of a blog entry, in which case this kind of digression is OK….)

The business book. The theme was one I’ve been thinking about a lot lately: focus. The exact opposite of this blog entry, in other words. Focus.

The basic idea of the book was: To succeed at X, find the one thing essential to X, and focus obsessively on X, to the exclusion of all else. The key to success is not balance, but strategically and tactically appropriate imbalance.

Whoops!

I try really hard to focus. I really do. But there are just so many interesting things in the world. There are dozens of novels well-worked-out in my head, hundreds of musical compositions, hundreds sketches of theorems (70% or so of which are probably correct), three substantially different AGI designs (Novamente, plus one based on Hebbian neural nets, plus one based on automated theorem-proving), even a few movie scripts… Not to mention that outline-for-a-unified-physics-theory…. Egads!

I could get myself to focus 90% of my time on creating an AGI, and push philosophy, music and fiction-writing (the other intellectual/creative pursuits that are necessary for my existence) into the other 10%. But I don’t seem to be able to get myself to focus quite that fully on bioinformatics, or natural language software, or any other business opportunity with the potential to yield the money needed to fund the implementation of my AGI design. I’ve been giving Biomind maybe 60-70% of my focus lately (which is a lot, because I work an awful lot of hours each week compared to what most people consider “full time” – I don’t sleep a lot) – because it needs it -- and pushing AGI into the background, which is extremely painful to me emotionally and intellectually.

(I have no issues with focus in the micro-scale: when I work I work with total mental concentration no matter how much noise and chaos are going on around me and no matter what mood I’m in or how tired I am, etc. (Except when Zadi’s watching South Park on the TV next to my desk, as that tends to be funny enough to distract me…. The episode I just watched almost convinced me that I should give up bioinformatics and fund Novamente via recording a Christian Rock CD…. If Cartman did it, so can I! I like his algorithm: just take a love song from the radio and replace all occurrences of the words “you”, “baby”, “darling” etc. with the words “Jesus” or “Lord.” Try it yourself, it works surprisingly well.) The level of focus that worries me is, rather, the choice of which things to direct my highly-focused micro-attention to. Which is mainly a problem because what I really want to focus on isn’t what the world currently wishes to pay me to focus on, and due to having a family to support I have this irritating ongoing need for money…. Leading to difficult temporal-assignment-of-credit problems, such as how much time to spend actively working toward AGI, versus working on things-I-like-but-don’t-love (bioinformatics, at the moment) that may yield money to pay for AGI research in a couple years, versus things that put me in a peaceful and creative state of mind (music! weird fiction!) so that my work on things-I-don’t-love is more effective, etc….)

If it’s true that sustained narrow-focus is the prerequisite of success, this would certainly explain why the most successful people aren’t generally the most interesting ones. Balance and breadth tend to make people interesting to interact with on a sustained basis. People narrowly obsessed with one thing tend to get tiring quickly – though they can be exciting and fascinating to talk to for brief periods of time. My close friends tend to be broad and balanced people, yet the people I admire most often have more of a narrow-focusing nature.

Now I’m sitting on the airplane – had to stop typing for a few minutes to board the plane, and then wait until the plane was aloft to bring out the laptop, because of the peculiar urban legend (embraced by the FAA) that laptops interfere with airplanes’ navigation equipment. While the plane was taking off I decided to continue the theme of my morning’s reading, and I read a couple articles in a free onboard copy of “Harvard Business Review.” (Also a delightful article on hats in “Ebony,” but I’ll spare you the details of that one….”Make no mistake, it takes a certain amount of bravado to wear a hat. .... It’s like the exclamation point to a fashion statement. … Hats hint at the essence of the wearer, giving a peek into the soul of the Brother underneath….” Ah, humanity! Gotta love it!) The current issue of HBR contains an interview with Garry Kasparov, the recently-retired world chess champion, on the relationship between chess and business.

Amusingly, Kasparov had something to say about focus, in the context of his chess battle with computer program Deep Blue in 1996-1997. He reckoned the contest had been an unfair one, since Deep Blue was trained on transcripts of his prior chess games, whereas all transcripts of Deep Blue’s play were kept secret from him. He also said he thought Deep Blue couldn’t beat him on his best day. But he said he thought one of the big advantages computers had over human chess players was their ability to focus exclusively and narrowly. “Human players have to cope with a lot of external pressures and distractions: you have a family, you write books, you give lectures, you get headaches, you have to earn money. There’s a lot of stuff filling up your brain while you’re playing. A machine, on the other hand, is completely without distractions. This shows the weakness, the shortcomings of the mortal mind, which is a daunting lesson for human beings. We just can’t play with the same consistency as a computer. So it’s all the more fortunate that we have our intuitions to help us play better.”

Kasparov obviously spent most of his life narrow-focusing on chess. Yet, he remains a bit jealous of a computer program’s ability to narrow-focus even more intensively.

And it’s interesting to observe that, for a chess master, Kasparov is an unusually breadth-oriented guy. His style is that of a strategic risk-taker, as opposed to that of his arch-enemy Karpov, who was always more conservative and analytical. Kasparov likes to think about business, literature, politics, and human nature in general – as he says, “I do not like details.” Of course, to become world chess champion he must have learned an awful lot of details – but what made him a master was not merely his mastery of details; it was his mastery of details combined with a truly rare and powerful intuition.

Kasparov’s style of chess could only be conducted by a mind with some breadth as well as narrow-focus, because it relies on general principles and intuitions regarding strategy – principles and intuitions going beyond chess and applicable to other domains as well. On the other hand Karpov’s style of chess was more suited to a purely narrow-focused approach.

AGI, I suspect, is really only susceptible to a Kasparov-style approach -- or really, to an approach that’s even more breadth-centric than Kasparov’s. This may be one of the reasons why AGI is so hard. If achieving anything substantial requires narrow-focus, then how is it possible for anyone to achieve something that by its nature can only be comprehended and mastered by someone with tremendous breadth? Tres dificil, nyet?

Physical sciences and mathematics don’t generally have this property – a very hard problem like creating a relativistic theory of gravity (solved by Einstein a long way back) or unifying gravitational and quantum physics (not solved yet) is nevertheless defined within a fairly delimited formal domain, and can plausibly be solved by a mind narrowly focused on that domain. To do what Newton did, on the other hand, clearly required breadth combined with focus. He had to focus to solve the hard technical problems, but he also had to have a lot of breadth to figure out what were the right questions to address, drawing from the incoherent mess of concepts and ideas that was pre-Newtonian physics. The analogy isn’t perfect nor original but I guess it’s an OK one: the task of creating AGI seems roughly comparable in magnitude to the task of creating Newtonian physics. Both have a conceptual and a technical aspect, though in Newton’s case the technical aspect was mainly mathematical, whereas in the AGI case it involves software design and engineering as much as mathematics.

Newton made his biggest breakthroughs during a three-year period when he was largely isolated in his house, at a time when England was mostly shut down due to the bubonic plague. (And, according to my university philosophy professor, his dog was named “Diamond.”) Maybe that’s what I need right now – a dog named Diamond, and an outbreak of plague to hit Washington, forcing me to sit in my house isolated for three years and do nothing but work on AGI by myself. Of course, the plague would have to hit the Internet too – isolation is harder to come by these days. Nah, that’s just a silly thought – software engineering, unlike mathematics, is better done by a small “extreme programming” style team than by a single individual. Plus, I don’t quite trust myself to teach a baby AI alone; the baby needs a woman as well as a man as a teacher (Izabela, with some help from Zadi?) and it needs a strong dose of Cassio’s conservatism and good judgment. What I need is for the plague to strike when I’m stuck in a house with the 4 or 5 best members of the Novamente team. And preferably it’s a big house, so there’s room for my kids and dogs with their noise and chaos in a separate soundproofed wing! (Yeah, yeah, this is just a stupid joky digression, please don’t quote me as if I seriously want a plague to come down on the world, I don’t…. (I started thinking that, since I happen to live in the Washington DC area, a plague in my local region might end up having some positive effects due to eliminating a lot of politicians. But, the body politic seems to have a self-regenerating characteristic similar to the limbs of certain lizards. And of course, a plague here in DC would probably be mistaken for a terrorist attack, which might cause Dubya to annihilate the continent of Africa by mistake or something … OK OK, enough!)

Wow, this is a long blog entry! I’d better call it to an end now. I’d intended to spend this flight finalizing the manuscript of “The Path to Posthumanity” – which, I recently noticed, the publisher has listed on amazon.com, even though I have not yet actually sent him the text of book! Well, some things move fast these days. Unfortunately I’m not going to be able to make that book nearly as good as I’d like, due to lack of time rather than lack of ability. I need to get that one out of the way so I can get back to finishing “Foundations of Emergent Cognition” (the shiny new name for the “Novamente book”), which is pretty much done, but just needs a final going-over, addition of references, clarification of which aspects of the discussion pertain to Novamente in particular versus pertaining to “any sensible AGI design,” etc. etc. Still, maybe “Path” will get some sales riding on the coattails of Kurzweil’s “Singularity.” The books cover much of the same ground, but mine gives fewer exponential and hyperexponential charts and more scientific depth – and mine also gives a more transcensionist, less “kinder, gentler” view of the Singularity. (Kurzweil is brilliantly insightful, yet he often seems to downplay the dramatic nature of the Singularity even as he trumpets its inevitability. Sometimes it seems like he foresees a Singularity full of modified or uploaded humans with shiny new gadgets – rather than a fundamental overthrow of the current order of mental, physical and social being. Of course, we may well get BOTH of these, but it seems a bit disingenuous to focus primarily on the former, even though it’s easier to understand and goes down better on Main Street. But of course, these comments are based on not having seen his book, which hasn’t been released yet – they’re based on his prior books and his speeches and online writings – maybe his book will give fair time to the transcensionist aspects as well, we’ll see.)

Enough – enough rambling, Ben – enough. Focus! Focus! Focus! Finish “Path to Posthumanity” and send it to the damn publisher! Write those Biomind press releases! Test the new Biomind ArrayGenius release! Finish the Novamente book! Launch the damn Singularity already so you can give yourself a better temporal-assignment-of-credit algorithm, eat 7 cakes without gaining weight, and push your daughter on the swing and canoe past crocodiles while composing weird jazz fission and programming meta-Haskell and kiss your wife while proving theorems that are themselves hyperdimensional conscious beings… yadda yadda… Focus! Focus! Focus!

Ah – wait – one more afterthought about focus. I had a Buddhist friend once who, every time I made a mechanical mishap like spilling a cup of tea, would point out to me: “See, if you were an enlightened Buddhist master, you’d never do anything like that. You’d never spill your tea because you’d be totally focused on whatever you were doing, in that moment!” In fact, this guy was neither particularly enlightened nor particularly focused nor emotionally balanced himself, though he was highly adept at pointing out the unenlightenment of others -- but he did have a point there. But of course, my retort was always “Fine, but I don’t WANT to focus my total attention on something boring like holding a cup. I’ll accept a certain error rate with boring things in order to focus most of my attention on interesting things. It’s no wonder no Buddhist master has ever achieved anything fascinating in science or mathematics – these things require focus in themselves, which is hard to obtain if one is focusing all one’s attention on drinking tea or raking leaves or breathing!” I think the analogy between Buddhist mindfulness and narrow-focus-for-business-success is not totally spurious. (Yeah, this brief paragraph doesn’t come near doing justice to my thoughts on Buddhism, but that’ll be saved for a later blog, it’s a deep and complicated issue in spite of its perfect simplicity, yadda yadda.) One problem is that the human mind is so painfully limited that it’s hard for it to do even one thing well, and when it divides its attention, it’s bound to make mistakes. Another problem is that we were probably evolved to focus on one intensive task at a time – like hunting, or escaping, or mating – and the modern emphasis on multitasking (on various time-scales) is an abuse of our evolutionary neural and physiological wiring.

Enough, OK, OK. Focus! Focus! Focus!

(I spent the flight from Salt Lake to San Fran sitting next to a very intelligent mining engineering executive who spoke very passionately about nutraceuticals and was bringing in a couple hundred thou a year selling them, via a variant of the classic “multilevel marketing” scheme. The nutraceutical line he was hawking actually seemed decent – founded on reasonable science – and I was almost convinced to give up the idea of making money for AGI through narrow-AI businesses and make the money through selling skin lotions and nutritional supplements instead. It might be a lot easier. I almost followed that plan when I was 13 and now I’m sorta wishing I had. OK, not really. But it’s an amusing thought…. I’m not such a bad salesman if I’m selling something I believe in; if I were selling e.g. life extension oriented supplements with some foundation in biology, I could probably give a convincing rap. My wife knows a lot of vain women in Brazil; maybe we could start the business in Brazil…. I’ve often thought that mixing up making money with AI is a mistake – it might be better just to keep my AI work pure and just accept that I need to spend a percentage of my time on some stupid business in order to pay the bills and hopefully eventually make enough money to pay the Novamente team to actually work on AGI engineering and teaching….. But, yah yah, the problem is that making any business work takes a lot of focus and attention, and it’s hard for me to see myself getting motivated to direct much of my focus and attention to something so boring as selling skin lotions…. The marketing ploy is slighty clever though: they suck women in my selling them skin lotion, and then upsell them to more expensive nutraceuticals, pointing out (correctly) that the key to beautiful skin is good health. Well, this sort of shit is what most humans seem to be interested, right? Beautiful skin, big muscles, good hair, shiny teeth and symmetrical faces. If you can’t have them yourself at least you can watch them on TV! (OK, OK, I’m not really going to quit the AI business to sell skin lotion. Although I’m not sure it would be a stupid idea in the medium term; in the short term I don’t have the stomach for it…. And anyway Biomind’s business prospects are actually looking pretty good right now (sales pitch ahead: anyone want to buy some of the world’s best microarray data analysis software?)

Foooocuuuuusssssss…...

8 comments:

Marc_Geddes said...

Fascinating, fascinating.

Focus is one the themes in the Matrix trilogy of course.

'Focus Trinity'!

'Don't think you are, *know* you are' !

and so on

Bob Mottram said...

That's quite a ramble. Reading some of your stuff (I have a PDF eBook of yours that somebody gave me on CD) I couldn't quite decide whether you were just a crank or not.

I liked the Newton's dog anecdote, and the advice from Buddist tea drinkers.

I'm not quite sure whether I believe all that signularity stuff or not. Opionions on what "the singularity" is seem to differ somewhat. Perhaps it's the point at which Kurzweil's grocery bill begins to rise exponentially. Whatever the real definition may be I'm fairly sure that it's not going to happen any time soon (like, in 5-10 years).

I can sympethise with your AI dilemma. Obviously we would all like to be doing AI research/programming all of the time, but there is the is the inconvenient difficulty of trying to persuade someone to pay you for doing it.

Ben said...

Bob Mottram says he doesn't thing the Singularity is likely to come in the next 5-10 years. I agree, although I'm not 100% certain it won't.

He also muses on whether I am a "crank" or not.

According to Wikipedia

en.wikipedia.org/wiki/Crank_(person)

"
Crank (or kook, crackpot, or quack) is a pejorative term for a person who writes or speaks in an authoritative fashion about a particular subject, often in science, but is alleged to have false or even ludicrous beliefs. Usage of the label is often subjective, with proponents of competing theories labeling their opponents cranks, but typically is used to describe someone who is well out of mainstream opinion on a matter. In most cases the people labeled as crackpots turn out to be wrong.
"

According to this I suppose I must be a crank, because I speak about scientific subjects and am "alleged to have false or even ludicrous beliefs." However, by this definition all evolutionary biologists (for instance) are cranks, since they are alleged (by creationists) to have false and ludicrous beliefs....

On the other hand, my ideas about AI have been repeatedly published in peer-reviewed books and journals, so in that sense I'm not a "crank." Also I've held many university faculty positions; and my narrow-AI software for bioinformatics and language processing has been used by a number of government agencies and universities. So, sociologically, I'm on the fringes of, rather than fully outside, the standard academic/commercial AI community.

Of course, though, me not being a "crank" doesn't make my ideas about AGI correct!

BTW Bob, I looked up your humanoid robot project and I find it quite interesting. Maybe we will work together someday, using Novamente to supply the cognition underlying your robot body....

-- Ben

caspar said...

the last thing you'd want is a dog like Diamond..

apparently the playful mongrel knocked over a candle, setting fire to several extremely important manuscripts that contained years of work.

"O Diamond, Diamond, thou little knowest the damage thou hast done."

Bob Mottram said...

I think the life and times of Diamond would make a fun work of fiction.

I wouldn't get too worried about being a crank or not. Plenty of people have called me one at various points.

I wouldn't be against the idea of a colaboration with anyone if I thought they could make a useful contribution. The few times that I've tried before have been pretty unsuccessful, mainly because there are very few people with any practical (non-armchair) AI experience. Gettings robots to perceive things or interact with them is a very down to earth business.

Anonymous said...

Pity you never expanded on the reason why, of the three substantially different AGI designs (Novamente, plus one based on Hebbian neural nets, plus one based on automated theorem-proving) you favoured the novamente one. Perhaps some future blog. Keep writing - I love reading them.

xuruargo said...

Couldnt agree more with that, very attractive article

online pharmacy

kuwuwait said...

Super-Duper site! I am loving it!! Will come back again รข€“ taking your RSS feeds also, Thanks.

nolvadex