Andy Brett is an engineer, writer, and athlete

The third James Bond movie, Goldfinger, opens with an amphibious mission to destroy an illicit chemical processing facility. After emerging from the water and planting the explosives, Bond strips off his drysuit to reveal a perfectly pressed white tuxedo and calmly affixes a red carnation to the lapel.

It’s rare that the next step after “take off your wetsuit” is “attach carnation to lapel” — it’s not routine, even ignoring the exploding chemicals bit. The Bond franchise and other filmmakers succeed by mastering this device, by leading viewers along a path and then suddenly yanking them out of the context they thought they were in. In some cases it’s overused to the point of becoming a trope (and ironically, somewhat predictable).

Another example is natural language processing — what Google uses to guess what you’re searching for, like it did just now when I typed “natural la”. In most sentences that start with “he left the keys on the kitchen…”, the next word is “table”, or maybe “counter”.

The drysuit-to-tuxedo reveal is the word that you don’t see coming. It’s the equivalent of completing the sentence above with “ceiling” instead of “table”. When you read a sentence or watch James Bond, your brain automatically produces guesses and predictions for what’s going to happen next.

When our guesses are wrong, we become engaged, or upset, or we laugh. We feel alive. That’s because bucking routine, and the expected, is uniquely human — for the moment, anyway. Artificial intelligence bots like Watson or competitors in the Turing test are usually confounded by the absurd or unexpected. It’s conceivable that they could eventually do quite well with it though, especially given how quickly humans turn the absurd into the hackneyed.

Avoiding routine

Falling into a routine can diminish your potential, not to mention your sense of free will. Athletes know that if they maintain constant mileage, weight, or intensity for a long period of time, their muscles will eventually stagnate. Much better is constant growth, followed by recovery.

How can you avoid falling into a routine? It’s helpful to have an outsider’s perspective. If you’re the one that’s stuck in the rut, it can be hard to notice and break out. The path of least resistance is usually the one that doesn’t bend much from the current trajectory.

There are plenty of hacks to get around this. The oldest, low-tech approach is regularly talking to a good friend who is willing to act as a sounding board and give candid feedback. A more formal approach looks something like a life coach — regular checkins on stated goals and aspirations.

In the life coach situation you’re paying, often dearly, for personal attention from a real human being in order to keep you on your toes. But what you get from a life coach (or any outsider, really) isn’t all that different from what you might tell yourself to do. If only you could have your present self communicate effectively with your future, soon-to-be-present self and guide him through evaluating some things in the cold light of yesterday.

Even before we get AI bots that can understand subtle, even British, humor, software is eating the world[1] in this domain, all the way at the top of Maslow’s pyramid.

That may seem like a ridiculous conclusion, and contrary to everything I just said — that you could write software to make people more spontaneous, more human, and more fulfilled. I suspect, though, that this falls into the category of problems where the right combination of AI and UI proves to be incredibly powerful.

Notes

[1] I made a first attempt at an app that would add spontaneity to your day, with Whimsical — every day you get a new challenge to complete that will likely force you to break your normal routine. There’s also Everest, which is focused more specifically on goals that you set up for yourself and capturing “moments” as you progress towards them. Sort of a self-directed life coach. And of course there’s Beeminder, which is the ultimate way for your present self to get the attention of your soon-to-be-present self — through his wallet. Some kind of meta-beeminding goal could do the trick nicely.

Lance Armstrong taped an interview Tuesday to admit that he used banned substances when he won his 7 Tour de France titles in 1999 through 2005. In 2001 he taped a Nike commercial with the voiceover:

This is my body, and I can do whatever I want to it. I can push it; study it; tweak it; listen to it. Everybody wants to know what I’m on. What am I on? I’m on my bike, busting my ass six hours a day. What are you on?

Most people couldn’t finish a Tour de France stage if they were “on” a bike with an engine. That Armstrong was on banned drugs doesn’t negate the grueling work he put in just to get close enough where the medicine put him over the top. I can certainly imagine how sitting in a hospital bed, near death, and watching as all manner of chemicals are injected into your arm might make you very comfortable with the idea of “substance X goes in and has effect Y.”

But that’s not the point.

Lopsided progress

Paul Graham in “The Acceleration of Addictiveness”:

It’s the same process that cures diseases: technological progress. Technological progress means making things do more of what we want. When the thing we want is something we want to want, we consider technological progress good. […] When progress concentrates something we don’t want to want — when it transforms opium into heroin — it seems bad. But it’s the same process at work.

Progress also yields EPO, and all of the other incredible substances that can alter the body and the brain. It would be impossible to develop the substances safely without developing incredible techniques in parallel for carefully studying their effects.

We usually only study carefully when we already know something abnormal is going on: testing a drug, or monitoring recovery from surgery. Like taking your car to a mechanic, there are a bevy of tools to diagnose any problems and just check on how things are running.

But once the car or human is back on the road, it’s back to operating with very limited instrumentation.

A car’s dashboard gives a readout of things like speed, gas tank level, engine temperature. These are roughly equivalent to the brain’s ability to convey things like hunger, thirst, or exhaustion. But in fact the car’s dashboard is more objective. It reports the actual levels, instead of just yelling “I want more” like a petulant child.

Let’s add the petulant child to our analogy – as the driver of the car. Your rational brain is mainly just making suggestions from the back seat, and can’t see the dashboard directly. He gets occasional reports on the dashboard from the child, who may or may not be telling the truth.

There are many ways to deal with a petulant child. You can guide him away from situations where he’s going to demand appeasement (not buying low quality but highly appealing food in the first place). You can buy earplugs that soften his screams (taking ibuprofen after skiing for 9 hours).

But what’s the equivalent of telling the child to “use your words”? Making him tell you why he’s asking you to do this? Or making him tell you the truth about what he can see on the dashboard?

This is where the progress becomes lopsided. There are all sorts of ways (drugs) to cause very precise and potent effects in the body. We have had ways of dealing with the petulant child, like ibuprofen, for decades. But we’re still only capable of muffling his screams or distracting him.

Our instrumentation, effectively, still sucks.

When you get a signal for “I’m hungry”, wouldn’t it be better to know why you were getting it? Are you getting it because something just reminded you of food, or are you bored and listening more closely for the signal, or are you actually low on reserves? What type of reserves are you low on? The human body isn’t just powered by a single chemical. It’s more like a hybrid car, except it can use not two but several different types of energy stores. Is my brain getting hungry signals because I’m low on glycogen? Maybe that’s fine if there’s a bunch of fat stored up and my rational brain knows that it actually wouldn’t hurt to use those reserves for a while.

People are more likely to be “on” a bad diet, or a new exercise routine, or too little sleep, than any serious drug. The effects of being on these more subtle “drugs” are far harder for the conscious brain to spot, and easier to ignore. Am I feeling lethargic around 3pm because I went with street meat for lunch? Ran 14 miles yesterday? Went to bed late two nights ago? Maybe it’s because I didn’t eat breakfast, or because I had coffee today and now I’m crashing, or because I didn’t have coffee today and I’ve become dependent on it. Or maybe it’s because I just read a pop science piece about how people often slump around this time, and everything else is perfectly normal.

So what am I on? I have no idea. But I sure would like to know, and it seems like getting this petulant child to grow up is a pretty interesting problem to solve.

The subway stops are numbered. The cars are numbered. The exits on the cars are numbered. The *doors on the exits* are labeled “A” and “B”. That last one might not do anyone any good except for the people doing maintenance on the cars, but the other labels sure do. There’s a sign in every station that tells you which car you should board based on the stop where you’re going or line you’re transferring to. In NYC there’s an app (Exit Strategy) and an entire subculture based on figuring that out. Tokyo just does it.

No one ever runs for the closing doors on the subway. That’s because the next train will arrive in about two minutes. This isn’t like the 1 train in New York where the conductor says “there is another train right behind this one” just so people don’t jam the doors when in fact the next train is 10 stops away and he knows it. No, the next train is always three minutes away, or less. It’s magical.

Speaking of magical, there is a magical card, the PASMO/Suica, that lets you ride *any* of the trains in Japan without having to buy separate tickets. This is the equivalent of the NYC subways and buses, NJ transit, the LIRR, the PATH, the Metro north, the T in Boston, and the Amtrak all accepting the same form of one-tap payment. It’s also accepted at vending machines and a lot of convenience stores, where you also just tap it against the card reader to pay. Not to mention the card’s mascot, which is a hilarious penguin.

There are machines to reload your penguin card *after* you enter the gate to the station, i.e. the most perfectly logical place to put one since it’s the place where you have free time. This seems really basic but SF and NYC both don’t do it. It’s good that it only takes about 30 seconds to add value to the card, because (see above) your train is arriving in the next 90 seconds.

The subway map looks like a bowl of spaghetti. This seems like a downside at first, but what it actually means is that it’s really easy to get to exactly where you want to go, usually with at most one transfer. There are also about 40 different maps of the various train systems. Again, seems bad at first blush. But each one is geared towards a person in a particular frame of reference, so if you’re looking at the one you need, it’s actually great.

There is cellular coverage on all the trains. But you never hear anyone yelling into their phone. Do you know why? Because there are signs that say “please don’t talk on your phone.”

Similarly, there are very few trash cans in the stations, or anywhere really. Some vending machines have receptacles built in for the stuff they sell, but that’s about it. If this were the case in the US, there would just be trash all over the place. But there’s not. Because people know not to throw trash on the ground.

Even apart from trash, the trains are *clean*. Both the cars and the stations. There are padded seats and about one strap handle every 8 inches (at two different levels on some cars). And these are the regular cars – a lot of longer distance trains have “green cars,” the equivalent of first-class cars, where I assume you can eat soba noodles off the strap handles. I actually noticed – and felt bad – when the condensation on a cold water bottle I was carrying started to drip on the floor.

Updated Sunday, March 11, 2012

A couple weeks ago I started toying around with Jekyll, an open-source, “blog-aware” static site generator. I really liked the idea of writing post content in Markdown with TextMate, and storing it in a git repository instead of a database. I got my old posts imported and transformed from html to Markdown, and pushed the site to GitHub. However, I also wanted to have images as part of the site, which seemed outside the scope of what Jekyll was designed for, so I waited to actually make the switch.

To be honest, though, the image capture/storage experience on my previous hand-rolled solution was never really all that great. It was cumbersome to upload photos on mobile, and you couldn’t see/manipulate previously uploaded images very well.

This got me thinking – what if I could just drop an image into a public Dropbox folder and have that publish it to my site? Dropbox has already made it easy to upload photos via their mobile app, so no need to reinvent the wheel there. Just like Jekyll has made it really easy to create and publish words, using Dropbox would make it really easy to publish photos.

The ideal solution would really be if Dropbox provided a way to just embed a photo gallery from a public folder. But alas – even with their shiny new release this week, there isn’t an easy way to do this.

So this morning I set out to kludge together a solution of my own. Using the ruby dropbox SDK gem, I have a script that pings Dropbox to see if any photos have been added to the “Public/Photos” folder (first checking the returned hash from Dropbox, and then checking each path for uniqueness against a local redis instance). If it finds anything new, it grabs the public link and uses that to create a new post in my Jekyll install. Then, using the ruby-git gem, it commits the new file and pushes to GitHub. When GitHub receives the push it regenerates the site with the new photo page.

Pretty Rube-Goldbergian if I say so myself. For all the gory details, or if you’d like to try this yourself, the code is on GitHub. I’ll be making various improvements as I have time. Pull requests welcome.

About a month ago I wrote about some of the issues I had with the Jawbone UP. I’m still using the UP, and happy to report that I’ve fixed the last item on the list – the lack of a web interface and data export.

It’s based on the unofficial api for the UP. When I saw this I published a little ruby gem for it and used that as the basis for an app.

I’ve published the (very primitive) app at jawbone.heroku.com. You can go there to see some demo data from my account.

If you put in your login credentials for jawbone.com in the upper left there, you should see your data displayed instead of the demo. You can also then go to jawbone.heroku.com/data.json to download it.

If anyone does try it out, let me know on twitter or email. I have a lot of ideas for how to improve this further.

Note: the app isn’t storing anything, including your login information. It literally does not have a database attached, so it’d be impossible to do so. Once you enter your login information the app gets a token that allows it to retrieve data from Jawbone and stores that token in a cookie. To “log out” you can just clear the cookie.

When I was younger I liked to read books about kids who were detectives, like Encyclopedia Brown. There was another series about a girl named Cam Jansen, who was aided in her detective work by her photographic memory. When she wanted to remember a scene, she simply said, “Click!” and it was permanently preserved.

Soon, anyone will be able to do this.

I have a feeling that Path realized this a while ago and is skating to where the puck is going to be. Their cutesy demo video shows seamless transitions between real life moments and the snapshots being captured in Path – in nearly every situation it would have been incredibly awkward to say “hold on, let me take out my camera to get a shot of this.” Path is executing on the ideal model for this future world in that these moments are “private by default”. No data is mined, no social graphs are monetized.[1]

There are several other situations where some sort of wearable device vastly improves an existing use case. I encounter one almost every time I go running – inevitably I’ll want to capture some interesting piece of scenery, but I never bring my phone with me, and even if I did I wouldn’t really want to stop to take it out. Put the camera on my sunglasses – boom. Problem solved.

Another use case in need of improvement is one of the “features” of the Jawbone UP – right now you’re supposed to take a picture of every meal you eat. Sites like Foodspotting have managed to get the early adopter set to make a habit out of photographing every dish before they dig in, but if it’s as effortless as saying “click,” suddenly it’s a lot easier to get people at risk for diabetes to keep a journal of what they’re eating.

Thanks to Sam Grossberg for unwittingly listening to a first draft of this post over a beer a couple weeks ago.

[1] Here is my plea to Path: Don’t let me be a free user. I know that I can buy filters and buy songs on iTunes via Path, but that’s not really tied to the value that I’m getting from the app. How about a voluntary subscription a la Instapaper?

There were actually two distinct themes at this year’s edition of Startup School, but the first makes for a better title. This is not a recap of everything that was said – for that, I believe you can watch the individual videos on justin.tv eventually.

Theme One: Do not, under any circumstances, start a startup.

This is, of course, tongue in cheek. But more than half the speakers said, in one form or another, do not start a startup. The summary went something like:

Don’t start a company just for the sake of starting a company. Don’t start a company because it’s cool. Don’t start one because you think it’s going to turn out like it does in the fictional movies, because it won’t – death is the default for startups. Don’t start a company because you want to be like the founders you see at startup school, or want to be the “next [fill in the blank].”

As one member of this chorus, Paul Graham offered one of the more analytical and level-headed critiques of the current landscape. He spent a few minutes before the Office Hours segment to give a mini-lecture on what he’s seeing with this round of YC applications – he described some of them as “what would happen if you did Markov chaining on a corpus of TechCrunch articles.” One startup that came on stage later was asked point blank, “Is this the biggest problem you could find?”

Paul noted that Bill Gates was trying to solve a problem he had when he wrote the BASIC interpreter for the Altair. He wasn’t thinking that it would lead to a giant software company or even a company at all. This was later echoed by Mark Zuckerberg who said that Facebook became a company in spite of his intentions – it got so much momentum behind it that it was going to happen whether he kept going with it or not.

In fact there was a palpable undercurrent of exasperation from a number of the speakers who had probably seen, heard about, or been pitched by one too many fresh-faced college students with billion-dollar ideas and an odd affinity for ramen noodles, hoodies, and caffeine. Ashton Kutcher, of all people, expressed this best when he said:

“I talk to these founders with these big ideas, and they jump from the problem they’re solving to who they’re going to be or what they’re going to get. I hear ‘It’s going to be a billion dollar company’ and I have this switch in my head that slowly shuts off. Because they’re jumping to the effect. You have to be the cause…If you want to be Mark Zuckerberg the best you’re going to be is second place. Because Mark Zuckerberg will always be a better Mark Zuckerberg than you.”

If Ashton’s distaste for founders trying to be the next Mark Zuckerberg by following exactly in Mark’s footsteps sounds familiar, it’s because Paul Carr wrote about it back in March[1]. Here’s the conversation Paul had with such a founder:

More worrying was his answer to the question “What are your plans for the app?”.

Without a flicker of irony, he replied “we’re going to be a billion dollar company.”

“You realise that’s not just a thing you can decide to be?” I asked.

A glare this time. A shrug. And then again, as if he were addressing a simpleton: “We’re going to be a billion dollar company.”

“How?” I said.

“You’ll see,” came the reply.

It was only then I noticed his outfit. Everyone else was in smart-ish jeans and shirts, but the entrepreneur was carefully dressed in a hoodie and a pair of open-toed flip flops. Later investigation would reveal that his “billion dollar” app was a social network for people with .edu addresses. The secret sauce? The fact that it gave college kids a way to flirt around campus.

Any of this sounding familiar?

“Doing a startup” has seeped into popular culture as a “cool” thing to do. It may come from Aaron Sorkin (Oliver Stone revisited), or it may come from the cadre of mainstream news outlets who don’t want to appear foolish by dismissing the next big thing and who therefore cover companies and founders earlier than ever. I think McSweeneys puts it pretty well with their update on the opening line of Jane Austen’s Pride and Prejudice:

“It is a truth universally acknowledged, that a single man in possession of a good fortune must be in want of an internet startup to call his own.”

So why should you start a company?

Because you’ve found your mission. When Mark Pincus took some time to think about what he wanted to do next, he asked himself what he wanted to be doing for the next decade (or three). He describes building a company as “building a house that you would want to live in” – the timeframe of building and owning a home was chosen deliberately, I’m sure. Many founders Mark talked to are willing to pursue something if someone else will fund it – they want to see if someone else thinks it’s a good idea. But that’s backwards, he says. You need to decide that it’s a good idea to be working on it first.

Work on a problem that you want to work on – but choose your subject wisely. If you’ve chosen correctly, the company will follow.

Theme Two: Design

The most overt reference to design was late in the day when Ron Conway launched a thousand tweets by saying that “Design and user experience is the new intellectual property.” There was a more subtle thread leading up to that. The very first speaker, Marc Andreessen, talked about the early days of Netscape when “nothing was instrumented” – echoing something Jack Dorsey said several months ago. Instrumentation isn’t what people normally think of when they think of design, but building systems that are transparent and convey important feedback to the right people in fact can require the most rigorous design discipline.

James Lindenbaum, one of the founders of Heroku, picked up the thread next when he discussed “Poka-yoke,” or preventing the user from making mistakes. To him, design is thinking through the emotional state of the user. He described Heroku apps as sitting in the middle of a huge ecosystem of services that, thanks to EC2, can ping each other in just milliseconds from around the world. Applications are being “decomposed” into services that they consume so that it’s completely clear how they can be valuable and solve new problems, rather than having to resolve old ones. This makes systems design and architecture essential.

Mark Zuckerberg also got in on the party. Jessica asked about the choices he made when scaling the company, which he described as just another engineering problem to solve. Getting a lot of teams working together is just another complex problem that you have to decompose and design a solution for.

It’s turtles – er, engineering and design – all the way down.

Drew Houston of Dropbox wrapped up the afternoon. Dropbox probably isn’t the first company you’d think of when it comes to design – their site is sparse and features mostly stick figures – but the whole point is that they’ve built something that actually works. Even the text file I’m using to write this is getting continuously backed up to Dropbox – seeing that little green check reminds me that I’ll have this text (and prior revisions) permanently.

It’s very easy to become jaded and have a cynical take on “startup advice” in general. To be honest when I walked out of the auditorium yesterday I thought I had come to the definitive conclusion that the content at the previous year’s startup school had been far superior, and that this year had been, to use the technical term, too platitude-y. But after letting the various talks percolate overnight I realized that the message just took a little longer to reveal itself.

[1] The irony here is that Ashton Kutcher is actually an investor in that founder’s company, but who am I to let the facts get in the way of a good story.