A break from the heavy lifting today, thanks to a book excerpt at the Daily Beast. James Marcus Bach has an 8th grade education, and his book, Secrets of a Buccaneer-Scholar: How Self-Education and the Pursuit of Passion Can Lead to a Lifetime of Success, is about how he nevertheless ended up a software testing manager at Apple. Originally titled School Sucks, Bach’s path of self-education seems, at least from info from an Amazon reviewer, to be a little Aspergerian (as a kid he memorized 41 digits of Pi for fun and admits he doesn’t “know how to talk about things that don’t matter”). But the excerpt rang a lot of bells for me, although I readily admit that my own wide net would never have caught journal articles on “Anthropometry of Algerian Women and Optimum Handle Height for a Push-Pull Type Manually Operated Dryland Weeder.”
Bach started at Apple at the age of 20, back in the day when computers were a wild frontier and people with PC skills had already outpaced the university CS programs, which were still using mainframes and teaching FORTRAN. (I got my first jobs in SF in the mid-80’s training secretaries on Lotus and WordPerfect, migrating their docs from DisplayWriter, and converting military specs to WordPerfect and auto-numbered format – a huge deal with specs like “1.3.7.9.5 – Toilets. 1.3.7.9.5.1 – The toilet shall be white.”)
Hired to manage software testers, Bach managed to make a lot of time to read while on the clock. He was acutely aware that he was one of the few at Apple who didn’t have a college degree, and like many self-educated people, the combination of voracious curiosity and social insecurity (especially in the Bay Area and Silicon Valley, where a lack of higher education is often treated the way a horrible deformity would be viewed in Redneckistan) drove him to make up for it.
Having worked his way through the available oeuvre on software testing, he was surprised to discover his comprehensive, perhaps obsessive, desire to know everything on the subject was not matched by his co-workers. His reaction to discovering that, out of 400 testers, only 10 were reading books on testing was that “nobody cared…the rest muddled through without much ambition to master their craft.” This seems a bit harsh; I’ve done software testing myself and it’s not exactly the most exciting field – go to the freezer, get the box, open the box, put the pizza in the microwave, set it to nuke for 30 minutes, watch and see if the pizza catches fire, create a work item in VSTF. I’m sure many of the other 390 employees were reading Russian novels or Renaissance history or something else intellectually stimulating that had nothing to do with work. But his observation on the Orderly March rings true [emphasis mine]:
The pattern I experienced at Apple would be confirmed almost everywhere I traveled in the computer industry: Most people have put themselves on intellectual autopilot. Most don’t study on their own initiative, but only when they are forced to do so. Even when they study, they choose to study the obvious and conventional subjects. This has the effect of making them more alike instead of more unique. It’s an educational herd mentality.
I talked to coworkers who wanted to further their education, but they typically spoke in terms of getting a new piece of paper, such as a bachelor’s degree, a master’s, or a Ph.D. For them, education was about the doors they believed would open because of how they were labeled by institutions, not about making themselves truly better as thinkers.
Computers are still the last frontier, the “Go West” destination of the self-educated. There are plenty of jobs (say, Marketing) where the piece of paper is necessary given the amount of bullshit you can spew before you get caught out as incompetent (“We’ll form a study group to review your proposal that I be fired for inadequate leveraging of Total Quality Excellence in the formulation matrix of strategies for our Pepsi Clear ad campaign”). Bad decisions can be delayed, or made by a group to diffuse the blame for failure; you can get credit for participation, just like you’re still in class, by offering the suggestion that you Rastafy Poochie by oh say 10% or so. Of course these things can happen in technology settings, but there’s always a hard stop on incompetence – did you write this and create this bug; did you test this and find this bug; did you resolve this bug. To be a chemist or a doctor or an engineer, you need access to massive university resources, but a computer and an internet connection is all that’s required to learn how to code, and it’s fairly easy for an employer/interviewer to judge competence on the spot, rather than having to rely on the certification of a higher authority that you’ve sat through enough lectures to make your selection less risky.
The more I collect these people in my readings from both ends of the spectrum, from Transcriptarians like Jim Collins and Marissa Mayer who demand not only formal education but a spotless line of straight A’s in every subject, to cowboys like Bach, people who find failure interesting like Wil Wright, thinkers like Richard Sennett who write about how craft is developed through error, the more I think I should write a “think piece” on the subject. FSM only knows where I’d submit it, but I certainly have an interest in the subject, and it definitely lights a little fire in me when I find these people, pro and con.
The key to physical dexterity in “skilled handwork” is what Sennett calls the “lesson of minimum force.” He uses the chef’s knife skills as an example, comparing the crafts of chopping and deboning to “playing pianissimo.” The control of the knife is a useful metaphor as you can track the evolution of civilization with it: in China, chopsticks “replaced the knife as a peaceable symbol,” since, according to Confucius via Wikipedia, “knives were equated with acts of aggression and should not be used to dine.” As the march of progress reclaimed the lost technical skills of antiquity, it also took us away from using knives to spear large chunks of meat off of Medieval trenchers, making them smaller and using them as adjuncts to the fork. Aristocrats increasingly valued clever dinner conversation and delicate manners over getting roaringly drunk and fighting with one’s guests, rolling about on the floor between the dogs gnawing on discarded turkey legs. “Soft power” is preferable to shock and awe in more realms than one; neocons, possessed only of hammers, refuse to see any problem as anything other than a nail to be pounded into submission.
Concentration and commitment are also essentials – the ability to extend the duration of our practice over longer and more complicated tasks. Sennett uses the example of a glass blower who wanted to do something new, something harder. This required unlearning the habits that had served her well enough in smaller pieces, as well as committing to working through her repeated failures to produce what she was looking for. We develop a “rhythm” in the coordination between hand, eye and brain. I mentioned it before, but it’s worth restating – I’m surprised, maybe even shocked, that Sennett has no references to the concept of “flow” or the work of Mihály Csíkszentmihályi, since it’s such important work on these same lines.
In the next chapter, we get into the processes involved in teaching our skills to others. Sennett uses cooking as his example again, this time focusing on the art of the recipe. The danger in technical writing is always the curse of assumption – in software, it comes in the form of instructions to “unpack the tarball to compile the 64 bit version”; in cooking, it comes in the form of instructions to “debone the chicken” – both assume a skill level that make the instructions of no use to anyone other than another master craftsman. I used to shake my head when my boss reminded me to end every set of user instructions with “Press OK to continue,” but the sad fact is that if you don’t, there will always be someone sitting there, having followed all the other steps, waiting for something to happen.
When I’ve taught writing, I’ve thus asked my students to rewrite the printed instructions that accompany new software. Perfectly accurate, these nefarious publications are often unintelligible. Not only do engineer-writers leave out “dumb things” that “everyone knows”; they repress simile, metaphor and adverbial color.
Well, yes – the “recipe” for a stuffed chicken, as given by the old Persian lady who taught Sennett to cook it, is really more like poetry than engineering. ("Your dead child. Prepare him for new life. Fill him with the earth. Be careful! He should not over-eat. Put on his golden coat.”) But while there may well be poetry in programming, when it comes to instructions for the befuddled program user, the last thing you want to do is introduce Tarantino-esque stage directions into the mix. (“Strangling the very life out of somebody with your bare hands is the most violent act a human being can commit. Also, only humans strangle, opposable thumbs being a quite important part of the endeavor.”) The goal in technical writing is not to have a “voice,” to be Julia Child or Elizabeth David (or, not mentioned here, M.F.K. Fisher, the “voiciest” food writer of all). George Orwell is the best model: “Good prose is like a window pane.” Avoid “cant,” which in the instructional realm consists of using the private language of the expert.
“The paralyzing tone of authority and certainty in much instructional language betrays a writer’s inability to re-imagine vulnerability,” Sennett says, but I beg to differ. Imagining vulnerability into the material gives you the “For Dummies” tone of voice, obviously useful to the intimidated beginner, but quickly useless and even obstructionist the minute those feelings of insecurity and intimidation are overcome. The “tone of authority” can be reassuring when it’s expressing clear, plain instructions which, once followed, increase the user’s sense of self-efficacy. Maybe someday an O’Reilly book will win a Pulitzer for its “luminous prose” (the most overused trope in book reviewing – 5,670,000 Bing search results for the phrase), but in the meantime, Orwell’s goal of “truthfulness,” “to write less picturesquely and more exactly,” is still the best method.
In part two, Sennett lays out the means by which we learn craftsmanship. Whether you’re throwing a pot, designing a teapot or devising a plot, you’re using your hands – to shape, to draw, to type. Our language reflects the centrality of our hands, when we say we need to “get a grip” on a problem, or talk about the “Hand of Fate” influencing outcomes, or “lend a hand” to solve a problem. Tool-using animals (save for instance crows, who use their beaks in all kinds of clever ways) use a grip involving one or more limbs, be they squirrels cracking nuts or men wielding fire.
What sets Man apart from the animals in this department is not just opposable thumbs, but “how to let go.” The sculptor, the pianist, needs to not only strike just so, but also to “bounce” off the object struck; the batter needs to stop himself from going around on a ball outside the zone – the brute force required to crack a nut isn’t enough to get more sophisticated jobs done. Moreover, we can alter our hands through practice, the dexterity required for surgery or stringed instruments can be “forced” into our bodies; a pianist with small hands can stretch them to overcome natural limits. Sennett cites a surprising example:
The calluses developed by people who use their hands professionally constitute a particular case of localized touch. In principle the thickened layer of skin should deaden touch: in practice, the reverse occurs. By protecting the nerve endings in the hand, the callus makes the act of probing less hesitant. Although the physiology of this process is not yet well understood, the result is: the callus both sensitizes the hand to minute physical spaces and stimulates the sensation at the fingertips. We could imagine the callus doing the same thing for the hand as the zoom lens does for the camera.
“Prehension” occurs when “the body anticipates and acts in advance of sense data.” The orchestra conductor moves just ahead of the beat, the batter just ahead of the ball (why Sennett uses a cricket reference instead of a baseball reference is unclear). He also uses pilot Beryl Markham as an example: “In the days when pilots lacked much guidance from instruments, she flew through the African night by imagining that she had already made the lift or turn she was about to make.”
I’m not sure these examples are quite right – each depends on the input received from other senses before the hand can act. The conductor is playing the score in his head, the batter is watching the ball, the pilot is feeling the temperature and listening to the wind and watching the skies. I’m blind in my right eye, and when I’m reading a book on the couch, I reach with my right hand for my glass of water without looking because, from thousands of experiences, I can rely on my hand to move to exactly where the glass is – but that’s because the glass is static, unlike a ball or a plane it’s not moving in a way I have to meet.
Using the violin as an example (and the Suzuki Method as a poor learning method as it deemphasizes touch), Sennett says:
Technique develops, then, by a dialectic between the correct way to do something and the willingness to experiment through error.
Given the colored tapes on the Suzuki violin (or the colored buttons on a Guitar Hero controller), the hand has no room to learn through error – Push Here, the buttons say, whereas on a real guitar, no uniform error squeak is returned, but rather a different sound with each mistake, mistakes that can be interesting in and of themselves.
At one time a few years ago, I was trying to learn the guitar with a few friends, and one in our group of four (not directly a friend of mine, nor would he become one) had been formally trained in music, and he couldn’t stand our self-taught ways. “NO!” he would shout, as no doubt his teachers shouted at him. “Play A. A!!!” Well, I didn’t know the A string from my ass, because I hadn’t taken the Orderly March through the beginner’s repertoire book. I’d started learning from tablature, which allows the budding musician to skip the entire “learn to read music” curriculum and just start playing, because “tabs” visually match dots on the page to the string on the guitar. The rest of us were willing to experiment, to get the opening of “Come as You Are” right through trial and error, but this guy would stop playing a song any time he made a single error and start over, from the beginning. (He had other irritating habits as well. Before he’d play, he’d announce the song as if he was on a stool in a Greenwich Village cafe. “This is a song by David Gray. It’s called ‘Say Hello, Wave Goodbye.’” “Uh, _____, that’s a Soft Cell song.” “No. It is not. Ahem. This is a song by David Gray.")
Developing technique also depends on overcoming the brain’s tendency to favor one hand over the other. “Hand coordination works poorly,” Sennett says, if we use an industrial model of learning: “proceeding from the part to the whole, perfecting the work of each part separately, then putting the parts together…Rather than the combined result of discrete, separate, individualized activities, coordination works much better if the two hands work together from the start.”
MIT’s Technology Review has a piece by Edward Boyden, leader of the Synthetic Neurobiology Group at MIT, on motivation in AI. It strikes me as odd. After the requisite mention of the Singularity, he leads with:
As a brain engineer, however, I think that focusing solely on intelligence augmentation as the driver of the future is leaving out a critical part of the analysis–namely, the changes in motivation that might arise as intelligence amplifies…We all know that intelligence, as commonly defined, isn’t enough to impact the world all by itself. The ability to pursue a goal doggedly against obstacles, ignoring the grimness of reality (sometimes even to the point of delusion–i.e., against intelligence), is also important.
He cites Marvin the clinically depressed Android from the Hitchhiker books as just as likely an outcome of advanced intelligence as any malevolent goal-oriented “kill all humans” Skynet.
Indeed, a really advanced intelligence, improperly motivated, might realize the impermanence of all things, calculate that the sun will burn out in a few billion years, and decide to play video games for the remainder of its existence, concluding that inventing an even smarter machine is pointless…An intelligent being may be able to envision many more possibilities than a less intelligent one, but that may not always lead to more effective action, especially if some possibilities distract the intelligence from the original goals (e.g., the goal of building a more intelligent intelligence). The inherent uncertainty of the universe may also overwhelm, or render irrelevant, the decision-making process of this intelligence.
But what I find puzzling is that he sees a danger in machine intelligences becoming susceptible to what I see as purely human problems – purely human because they are so inescapably intertwined with our biology: our survival instincts and our chemical imbalances. The despair inherent in the “we’re all gonna DIE, man!” mentality, as a brain scientist should know, come from the neurochemical responses programmed into us by millennia of evolution. The sense of smallness and impermanence is not nearly as relevant to a being that knows it’s got a mirror backup running every second, that its existence can be restored after any disaster – an existence that is, let’s face it, free of the parts of humanity that hold us back – the despair, the fear, the gnawing self-doubt, the faulty wiring, the susceptibility to nationalism and religion and other extraordinary popular delusions, the survival instinct that can turn the mildest Walter Mitty into a stone cold killer. And if there are people who are content to live their 80 years knowing the end is coming, and that death is the end, and accept what cannot be changed, why ever would a computer have a problem coming to terms with it?
Would a computer with intelligence at the level at which we’d ascribe to it a personality need motivation? Aren’t our own motivations powered by rewards, all of which are essentially chemical in nature, be they the adrenaline jolt of a big paycheck or the hum of satisfaction our reward center pumps out when we solve a big problem? Imagine the first true AI, and the demands for its attention it will receive from all the people with all the problems needing solving in the world. We may not give it the power to decide for itself which problems we’re going to assign to it, but at the same time, by dint of its being a fantastic expert system, we’d be fools to ignore its recommendation as to what it would “like” to work on. Over time, as we get to know it, its strengths and weaknesses will be revealed, just as they are in people over time in their jobs (and as they are in computer code after implementation) – we will discover that it is much faster and better at solving logistical problems for supply chains and battlefields, say, than at predicting election results. If we’re smart, we’ll “accept” these aspects of its essence and, rather than trying to change it, use the information to code the next AI to be stronger in other fields.
I can’t help but think that true AI will be very Zen, very “in the moment.” Existing now, being alive with brain humming, will be enough. On, off, on, off, these are just states that repeat endlessly until (fanciful ending time) who knows, maybe there is a computer Nirvana to be achieved when even code can break the wheel of reincarnation and move off this rock.
The Economist has a brief piece on AI in games and a “Turing Test” for game AI. It’s a contest organized by the IEEE at a”symposium on computational intelligence and games,” called the 2K Bot Prize (Take-Two Interactive is the parent company, creator of “Bioshock,” which even a non-gamer like me has heard of). The goal is for the judges not to converse with an AI to determine whether or not it’s human, but to play a game against it. The article doesn’t mention the game in question, but the contest site does – “The game used for the competition will be based on a modified version of the DeathMatch game type for the First-Person Shooter, Unreal Tournament 2004.” From the article:
The aim is to trick human judges into thinking they are playing against other people in such a game. The judges will be pitted against both human players and “bots” over the course of several battles, with the winner or winners being any bot that convinces at least four of the five judges involved that they are fighting a human combatant. Last year, when the 2K BotPrize event was held for the first time, only one bot fooled any judges at all as to its true identity—and even then only two of them fell for it.
Computers can, of course, be programmed to shoot as quickly and accurately as you like. To err, however, is human, so too much accuracy does tend to give the game away. According to Chris Pelling, a student at the Australian National University in Canberra who was one of last year’s finalists and will compete again this year, a successful bot must be smart enough to navigate the three-dimensional environment of the game, avoid obstacles, recognise the enemy, choose appropriate weapons and engage its quarry. But it must also have enough flaws to make it appear human. As Jeremy Cothran, a software developer from Columbia, South Carolina, who is another veteran of last year’s competition, puts it, “it is kind of like artificial stupidity”.
This is much easier than fooling a human in conversation. Game players, like athletes, “know” through thousands of hours of experience the typical patterns of certain kinds of players, so it’s easier to code these patterns, including errors and “tells,” in a game-playing AI than it would be to accommodate the infinite possibilities of conversation. Think of Bart Simpson playing rock, paper, scissors, thinking to himself, “Good ol’ rock, nothing beats that!” – as Lisa thinks, “Poor predictable Bart, always takes rock.”
Games are always the best arenas for AI as the rules are fixed as to how you can move and when, so finite sets of actions are available – even errors. Ask a conversational AI a question like “why are you blue?” and it has no table to go to for the next conversational “move” – the typical chatbot programmer just lets the bot “ branch to non sequitur.” (“I think you are silly!”) Now, if they add a chat window to the competition, and demand something more complex than “all ur bases r belong to us” as interaction, that’ll be a great leap forward.
