The only thing that really worried me about this job with Mylo was the fact that I’d be commuting across the lake again. When I was working for Microsoft, the grueling trek across 520 was a great way to ruin any day. Doing it on a motorcycle gave me just enough of an advantage to make it bearable, most of the time, but it was still pretty bad; on days I had to drive a car, it was so frustrating that – well, I don’t even want to talk about that, really. It was not a good part of my life.

I felt good the day I freed myself from that job and went roaring west down 520 for the last time. The sun was shining and the traffic clear, and I swore I’d never ever ever do it again, no matter how much money anyone offered me – money simply can’t buy back all the time I spent feeling frustrated and unhappy on that godforsaken highway.

Four years later, here I am again, working for a company on the Eastside, but this time I’m crossing via I-90 and not SR-520. The commute is actually longer this way – 12 miles instead of 10, and it was 10.5 to Microsoft – but so far it’s not half so unpleasant. The morning ride eastward is a breeze. 90 gets a little congested right at the beginning, but it only slows to 45 mph or so, and the nearly-empty HOV lane across Mercer Island gets me going up to 85 mph, or faster if I want. Whoo!! Going to work is actually fun!

There’s always a back-up on the ramp from 90 to 405, since 405 is always thoroughly clogged, but the bike makes it easy to squeak in at the last minute and then pop back out a couple hundred yards later. A quick triple lane change on 405 gets me into the HOV lane, then I’m back up to 60 mph for two miles, where a special HOV-only offramp dumps me onto NE 4th just a couple of blocks from the office. It is fully sweet and I really cannot complain.

The ride home is much worse, and I haven’t found a satisfactory route yet. There’s only two miles of 405 between the on-ramp from NE 8th and the intersection with I-90, but traffic crawls along so slowly that my speedometer doesn’t even register it as motion. After the mile of misery, there’s a mile of HOV lane leading to the I-90 offramp, but that’s slim consolation after sitting in the rain barely moving for 15-20 minutes…. The solution might be to skip 405 altogether and go straight down Bellevue Way to the I-90 on-ramp. It has stoplights and a low speed limit, but at least it moves.

I-90 westbound is great until the HOV lane peters out halfway across Mercer Island, dumping me back in with the rest of the traffic for the last 2 miles. This seems to be obnoxious but acceptable at around 5 pm, but it was fully terrible yesterday when I left around 6. Tonight I’ll try cheating my way past it by taking the HOV-only exit at 80th Ave, then re-entering via surface streets at the Mercer Way on-ramp.

We’ll see how I feel about this whole business after a few more months of rain, but for now I feel like I can handle it.

I observe that people at Mylo feel free to work from home – there are only two other people in the office today, and one only showed up midafternoon. I should be able to take a break from the commute any time it starts to feel oppressive.

postscript: Bellevue Way is no improvement: it clogs up at the intersection with 112th Ave, and the on-ramp to I-90 is metered, with no HOV bypass. Using Mercer Island surface streets to skip past the tunnel works great though.

On the “Do you want to be a programmer at fifty?” thing is an interesting entry in the discussion that seems to be sprawling over the tech-blog-o-sphere, but what really caught my attention was this:

James went on to identify two kinds of programming

Type A “work(ing) out the solutions to difficult problems. That takes careful thought, but it’s the same kind of thought a novelist uses to organize a story or to write dialog that rings true. That kind of problem-solving is satisfying, even fun.”

Type B “what most programming is about – trying to come up with a working solution in a problem domain that you don’t fully understand and don’t have time to understand… skimming great oceans of APIs that you could spend years studying and learning, but the market will have moved on by then … reading between the lines of documentation and guessing at how edge cases are handled and whether or not your assumptions will still hold true two months or two years from now.. the constant evolutionary changes that occur in the language definition, the compiler, the libraries, the application framework, and the underlying operating system, that all snowball together and keep you in maintenance mode instead of making real improvements.”

I hardly know what to say about this but I think it helps explain why I had such an unsatisfying time at Google: it was almost all type-B work, which to my way of thinking isn’t really programming at all. It’s just…. API-twiddling. Meaningless, brainless, unsatisfying. Who cares if it works, if you don’t know why, if you don’t understand what you built?

I’m sort of startled and disturbed that this would be considered “most programming” – that sounds like a terrible world to live in.

Skiers at night with EL wire and headlamps. So clear! I’m curious how they recorded this, because it’s crisp and gorgeous – little or no noise in the blacks, either.

Oh yeah. Winter is coming…

After wading through the incredible heap of nonsense it takes just to download the latest Xcode, I am seriously disenchanted with Apple. I never wanted anything to do with their iTunes-centric consumer-electronics universe, but now you have to sign up for an iTunes account just to download Xcode. I can’t even write code on my own Mac anymore without jumping through their hoops. Ugh. Ugh. When did they forget that it’s my computer not theirs? What, just because they wrote the OS, that means they get to decide what I can do with my own machine? No thank you very much go away now please.

Ten or fifteen years after the first Year of the Linux Desktop, it’s still a pretty rough experience, but at least they don’t drag you through their greedy corporate agenda at every opportunity.

If you are still using Snow Leopard, don’t upgrade: it’s really not very nice in the Lion world.

This is my first day at Mylo, a two-month-old startup which is going to do something with digital photography workflow management. After this afternoon’s all-hands meeting I’ve learned I’m not supposed to say more than that in public, but the pitch that got me on board was significantly more interesting.

The ride in to Bellevue this morning was about as pleasant as a commute ever gets: I even touched 85 mph at one point as I crossed Mercer Island. This at 9 AM, even! Perhaps I just got lucky today, but it could get a lot worse and still be a pretty reasonable trip.

This new Macbook with the Retina display is nice. I have sharper-than-average vision and have always been able to make out individual pixels, even with anti-aliasing, but with this screen the text might as well be perfectly smooth. It feels unreal, like it’s not actually a computer screen but some faked-out Hollywood movie prop.

Mountain Lion, though – ugh. It took almost two hours to track down and turn off all the annoying iOS-derived frippery and make it act something like a reasonable desktop machine. (Check out Lion Tweaks if you are interested in doing the same.) Seriously, people, if I wanted to use iOS, I’d be using an iPad. And I don’t, so I’m not, so please stop shoving that stuff down my throat, ok?

Oh, well, there’s nothing to be done about it. I don’t love the Mac OS anymore, but at least it doesn’t suck any worse than anything else. Apple just stopped caring about people like me after their consumer electronics business took off, and the tail has obviously been wagging the dog for some time now.

There is no electric vehicle charger in this building’s parking lot but there are chargers in the two adjacent complexes, and it’s possible they may install one here too. I think my electric motorcycle project is going to move back on the “active” list as soon as I’ve finished the chandelier.

Bellevue is a strange place. The building height distribution is bizarrely bimodal: there are the old one- and two-story suburban buildings, the background level of the whole area, and then there are the twenty-plus story highrises stacked up among them; there is almost nothing in between. I would guess that there are a total of maybe three buildings greater than two but fewer than twenty stories tall in all of Bellevue.

It’s obvious that someone is trying to build a city here from scratch, that the density has not developed organically as a function of demand. I wonder if this is a little bit like people must have felt in Dubai, as a mile of skyscrapers erupted in the center of what had been a sleepy little desert town? There’s no reason for there to be a city here, except that some rich dude decided he wanted to build one, and now you have sprawled-out suburban mall style buildings laying across the street from towering glass-and-steel highrises. It makes no sense, and yet here it all is.

Via Metafilter, a beautiful, moving, surprising collection of photos from a photographer named Christy Lee Rogers. Rich, intense, vivid colors and intense shadows, all thick and dreamy.

I have two projects right now; everything else is in cold-storage.

Most of my creative energy is going to the chandelier project. It is both complex and ambitious, so I’m running into a lot of obstacles; nothing is going as smoothly as I’d hoped it might. A month in, all I have to show for my effort is one test board, which sort of blinks but doesn’t really work yet, and a stack of sandblasted glass cylinders. Oh, well: part of the point was to stretch my skills, and I’m certainly doing that. This is a whole new level of electronics complexity, and there’s a lot of straight-up fabrication work, too.

I bought a drill press today: a 12-speed, heavy-duty, floor-standing drill press with a 3/4 horsepower motor. It’s a big old tank of a machine, likely as old as I am. I then built a jig out of plywood, ABS pipe, and a bucket. The goal is to bore a hole into the end of each of seven glass vases, which I can then use to hang them (upside down) and which will provide airflow to cool the electronics inside.

Once set up, I added a small fountain pump and a lot of water, for cooling. As I drill out the end of the glass, the fountain keeps fresh water circulating over the cutting area, and if I drill slowly enough the glass stays cool and doesn’t crack. It’s a slow process, but I hope I can have half of the cylinders done by the end of the day tomorrow.

The other project is my ongoing development of the Radian language. I don’t really talk about it here, because it has its own blog, but I’m still working hard and am increasingly pushing it up into “useful tool” territory. Recent work has focused on the string library, and I’m currently deep in Unicode territory building the uppercase, lowercase, and case-insensitive transformations.

I’m increasingly thinking about the human end of this problem: I’m now in the zone where advanced curious users could probably make something of the tool. In addition to all of its internal data management, it can read and write files and manipulate text – that’s a good start. Somehow I need to find the people who want this thing, and then I need to have a good experience ready when I manage to pique their curiosity: documentation, examples, a good installer. I think I need to decide what point I’m going to call “good enough”, in terms of basic built-in features, and when I reach it I should redirect my efforts toward documentation and recruitment.

I can stand to spend some time doing optimization work, anyway: there’s a lot of room for intelligence in the Radian semantic model which I have taken almost no time to exploit.

This is just the way things go sometimes, isn’t it?

Grrr.

Bikes saved my life

When something as random as a blood vessel bursting in your brain could kill you at any time, why waste energy trying to live life more safely? Safety is a myth. Let’s embrace life with all its risks, enjoy ourselves and really feel alive.

One critical precondition for successful urban cycling systems: get rid of the helmet laws.

“Pushing helmets really kills cycling and bike-sharing in particular because it promotes a sense of danger that just isn’t justified — in fact, cycling has many health benefits,” says Piet de Jong, a professor in the department of applied finance and actuarial studies at Macquarie University in Sydney. He studied the issue with mathematical modeling, and concludes that the benefits may outweigh the risks by 20 to 1.

He adds: “Statistically, if we wear helmets for cycling, maybe we should wear helmets when we climb ladders or get into a bath, because there are lots more injuries during those activities.”

I maintain that helmets are for extreme sports, activities where you are likely to injure yourself. If bicycling around a city is really that dangerous, there’s something deeply wrong with the city’s infrastructure – and if it’s not that dangerous, why require people to wear helmets?

I’ve never worn a bicycle helmet. I haven’t really ridden a bicycle since the helmet-law craze kicked in, either. Repealing the local helmet laws would go a long way to making me think of a bicycle as a normal mode of transportation, a machine I could use to run errands and visit my friends, instead of just a piece of equipment for an extreme sport I don’t happen to enjoy.

Of course I’m not talking about high speed cycling for sport, here – if you’re flying along at two-lane country roads at 30 mph a helmet might well be worthwhile. I’m just talking about the sort of “maybe I’ll take a bike to the store instead of driving” trips that would make us all healthier and less oil-dependent. We should be promoting that kind of thing instead of scaring people off by telling them it’s so dangerous they need a helmet so they won’t die.

Google’s Android file transfer app has an incredibly annoying habit of popping up every time you plug your phone in, whether you want it to or not, and there’s no preference to make it stop. This might be fine if you only ever plugged your phone in when you wanted to transfer pictures, but I use my Mac to charge the phone, too.

Here’s how to disable the behavior: it’s a bit finicky, but basically there’s a little “agent” app that does the auto-launch, and the Android utility app installs this thing as a login item every time you run it. So you have to delete the login item *and* rename the executable, thereby preventing the Android utility app from reinstalling the agent.

Summary, since that link will probably go away:
– If “Android File Transfer” is open, tell it to quit.
– Use Activity Monitor to force the process “Android File Transfer Agent” to quit.
– Right-click on the “Android File Transfer” application and select “Show Package Contents”. Go into Contents > Resources and rename “Android File Transfer Agent” to “Android File Transfer Agent_DISABLED” or something of that sort.
– Go to your user home directory and open Library > Google; if you see another copy of “Android File Transfer Agent”, rename it similarly.

You will still be able to open “Android File Transfer” by hand any time you want to look at the files on your phone: it just won’t pop up automatically every time you plug the phone in to charge.

Simtec Electronics offers a USB device called the Entropy Key: a PN-junction-breakdown based hardware random number generator.They’ve done what appears to be a solid job engineering this thing, with multiple independent entropy streams used to crosscheck the actual randomness of the output.

(Useful for SSH session keys and key-pair generation: basically it’s for secure cryptography.)

The SSM2166 is an integrated circuit which implements a single-channel preamp with adjustable compression and noise gating. Stick this on the input side of an ADC, in other words, to get a clean, predictable audio input.

I might be done with Burning Man.

There’s this thing with altered-reality experiences: they don’t last forever. You gain some new perspective through them, you acquire an expanded sense of what is going on in the world, but repeatedly pursuing those experiences doesn’t continue to expand your perspective. It has something to teach you, but once you’ve learned the lesson, what do you do with it?

I’ve been to nine burns, starting in 2001. The first was difficult, stressful, not a lot of fun, but it planted a lot of seeds. The next couple of burns changed my life and I would not be the person I am today without them. Those experiences stretched out my awareness of life possibilities and gave me a sense of purpose around the creative arts. Then there were a few years where I ran with that, throwing down the best I had to offer each summer.

Now… why am I going back? It seems to be a cycle of diminishing returns. As the event develops, it gets bigger, thus more civilized, thus more limited, and the sense of possibility I feel when I crunch across that old dry lakebed fades into memory. I know what I can do out there, and I’ve taken that back home and done my best to build a life where I can do that all the time. When I go to the burn now, more and more I see the scaffolding, the stage set, the limits of the illusion, and the less I feel any expansion of possibility.

Time to blow it all up and start over. Innovation becomes repetition becomes tradition. Time to strike out on my own. Floodland 2013.

I’m going to delay my re-entry into the workforce by another month to work on an LED project. This will be the most complex piece of lighting I’ve ever attempted, incorporating 1,344 individually addressable RGB LEDs driven with 24-bit gamma-corrected color. That’s over four thousand individual control channels which have to be timed at megahertz resolution. On top of that, the display will be running an evolving, adaptive, algorithmically-generated animation loop… and the computation will be distributed across a peer-to-peer processor network. I am so excited I can barely stand it.

Why am I doing this? My friends Michael T. and Likhita K. want a custom chandelier for their dining room and asked me to design it. We spent a few hours last week with a couple bottles of wine, kicking around ideas, and came up with a concept incorporating trees, waves, and fireflies… it’s going to be beautiful and unique, a little bit Art Nouveau, blown glass and wrought iron and braided copper… all powered by an array of ARM processors.

Formal specification of the grammar for POSIX regular expressions

Surprisingly difficult to locate in a search: most articles discussing “posix” and “regular expressions” and “grammar” discuss the use of regular expressions in parser implementation, where what I want to do is implement a parser for regular expressions themselves.

Back in Seattle, typing on my laptop at the local pub. I’ve had one night of sleep on a proper bed in a motel and another at home, but I’m still feeling pretty vacant. Aside from some desultory unloading and cleaning work, I’ve done almost nothing today.

Ava’s lungs really don’t seem to handle playa dust well, so she was pretty sick by the end of the week and not having a good time at all. I spent Saturday packing up the car, and we left straight after her burn-perimeter shift was over. A good steamy shower seems to have helped her; I hope the cough doesn’t linger for six months like it did last year.

I’m sorry to say it was an unsatisfying week, and I’d have done better to stay home. The event itself was pretty much the same thing as ever; I just didn’t have anything to do most of the time, and what I tried to do mostly didn’t work out. My friends were scattered all over the playa, thanks to the Great Ticketing Fiasco, and Ava was too sick to spend much time walking. I went out one night in a dust storm with the KAOS art car, and there was one other walking expedition which didn’t end in tears, but there were no “wow” experiences, nothing to make this year stand out, nothing that made me feel moved or inspired or changed.

Well, ok, there was a “wow” moment on Saturday night when they blew off the last round of fuel bombs in the man base, and the whole building erupted into an enormous fireball, and everyone covered their eyes and backed up as fast as they could because OH WOW IS THAT HOT. That put a grin on my face. It was a little echo of the titanic oil derrick explosion back in ’07.

All together, though, I didn’t have a good reason to go, wasn’t able to find any good reason to be there once I arrived, and I wish I’d stayed home. Oh well. With nine burns under my belt I guess they can’t all be awesome.

Tuesday morning I finished wiring up the playa scrambler. Midafternoon, I took a break to go visit Opscode for a couple hours. Then I came back and got down to business with what I thought was the last couple of hours needed to finish off the trike.

It’s a complicated machine, with three separate chain loops: one connects the motor to the rear axle, another connects the pedals to the front ratchet box, and the third, the longest, connects the front ratchet to the rear axle. Over and over, I’d get it all rigged up, take it for a test ride, exult in the thrill of its decent speed and astonishingly responsive steering, then stop with a sudden horrible noise and discover that the long third chain had popped off one or more of its sprockets.

I won’t bore you with the long, long list of hacks and kludges and clever tricks I tried out to solve each of the never-ending series of problems, but if you ever get a chance to look at the trike in person I’d be happy to show you. Some are obvious. I never got a test ride to last more than a block or two before failing. After working for eighteen hours straight, it was time for the trucks to leave; I hastily welded on one more chain guide, rolled the trike up onto the trailer and went to bed.

That lasted a couple of hours. Wednesday evening I came back to clean up the disaster I’d left in the shop, then I pretty much went to bed again, but not before emergency overnighting four small sprockets. Assuming they arrive during the morning on Friday and not the afternoon, I’ll take them with me and try to fix what will hopefully be the last set of problems on-playa.

Today I got up feeling about as wrecked as if I hadn’t slept. Ran some errands. Got a bit more than half of my “to do before leaving” items checked off. Spent two hours building a solar-powered phone charger out of bits from my parts box. It doesn’t actually seem to work and I can’t figure out why – all the components work individually… my brain is so fried I can’t really debug it properly. I’ll take it along anyway; maybe I can fix it once I get there.

and then… for some reason i decided to try to finish the playa coat project I abandoned when work on the trike started to get intense. I did not exactly succeed – the garment is missing its pockets and collar, the tailoring isn’t great, the inside is a mess of raw unfinished edges, the cuff stitching is laughably uneven, and the lower hem is just a quick pass with a serger – but it’s finished enough to keep me warm without looking too much like an unfinished project.

now it’s 1 AM. time to crash.

flight leaves 3:15 pm tomorrow. will I be ready? Of course I will. what choice do I have?

Motor controllers are hard to understand. It’s not that I can’t find information about them on the Internet, but that there is so much information and so little consensus. For example: can I use an Arduino + SSR to control a DC motor, or do I need to use a MOSFET? Do I need a capacitor? How big a capacitor do I need? Do I need a kickback diode if I’m using an SSR, or is that only important if I’m using a MOSFET? Why does this “simple motor controller” diagram have thirty components?

I am flexible on the control mechanism. I could get a motorcycle-style twist-grip throttle, but I could also just have a push-button and a knob, where the push-button means “go” and the knob means “how fast”.

Whatever it is, it has to be simple enough that I can feel confident it will work, and that it will survive the playa. I don’t trust these massive black box circuits full of who knows what.

Maybe I’ll just skip the whole controller idea and just give it a “go” button. Push the button to power the motor and thereby accelerate; release it to coast. Would that work? Maybe I’ll just try it.

I got the rear chainring mounted on the playa trike yesterday. It took all day: I wasted five hours laboriously designing and building an adapter plate that would allow me to bolt the chainring on to a set of studs welded on to the old sprocket, which in turn was bolted on to the wheel’s original studs, only to discover that this offset the wheel so far to one side that I’d have to cut away half of the swingarm to allow it to spin freely. Mounting the chainring on the other side wouldn’t work, because the motorcycle wheel hub is aluminum and thus not weldable. After that didn’t work out I was tired and frustrated so I tore it all apart, screwed the nuts on to the original studs, hacked out the middle of the chainring so it would fit over the wheel, and then just welded it on, maintenance be damned. It wobbles and it’ll be a total pain to replace, but whatever, it’s done and I can move on to the next thing.

Today’s project, assuming I can find a way to work through the amazing heat, will be to fabricate a bracket, mount the electric motor onto the swingarm, and chain it up to the largest of the rear sprockets. After that I need to rig up a motor controller. I have a little silver box with a lot of wires sticking out of it that is supposed to be the controller for the same model of scooter the motor came from, but I am dubious: it does not appear to be robust enough to survive the playa, and in any case it has too many wires for the number of functions I expect it to perform.

I’m thinking about rush-ordering a solid-state relay, pairing it up with a capacitor, a schottky diode, and a microcontroller, and calling it a day. I don’t need reverse and it’s easy enough to program in a fixed accel/decel curve as a simple low-pass filter on the input.

Now how to control it? The traditional solution is a potentiometer mounted in a twist-grip, but it occurs to me that there’s an existing input I could use instead: this machine has a set of pedals, after all. What if I measured the speed of pedal rotation and spun the electric motor proportionally? That is, you make the bike go faster by pedalling faster, and slow down by spinning them more slowly. This method of control might be wonky in the city, where you have hills to climb, but out on the dead-flat playa the only factor that matters is how quickly you want to travel.

This mechanism would also reinforce our claim that this machine is just an unusually elaborate e-bike and not an unusually small mutant vehicle, and therefore in no need of a mutant vehicle license. If you pedal it to make it move, after all, doesn’t that make it a bike?