The piece starts with the knocking of a woodblock which creates a kind of rhythmic gauntlet through which the orchestra has to pass. — John Adams
IN 1986, Michael Tilson Thomas and the San Francisco Symphony commissioned an orchestral work from John Adams for the Great Woods Festival.1 The resultant fanfare has become one of Adams’2 most popular pieces. In 2015, Adams’ work seems like a fitting anthem for the SF/SV startup scene, where many a hopeful donkey straps a horn to its nose and hopes for a myopic venture capitalist to mistake it for an undiscovered unicorn.3, 4
This piece came to mind a few Monday’s back, when the startup I was working for, Jut underwent a massive reorganization.5 A startup pivoting in San Francisco is fairly unremarkable (unless you work for it and the pivot involves laying off around 80% of the staff). On the other hand, when you were the last hire, 80% of the company is gone, and it was Day 11 for you, the event is a bit more noteworthy, and “Short Ride” takes on an entirely new meaning.6
Waiting in line for the Fast Machine
I applied to Jut as a result of a bad Google search. I was looking for a library to make plotting real-time data easier from an Arduino I had streaming JSONs from a sensor. I came across Jut.io, which had nice graphs from streaming data, but required a backend not appropriate for this application. Then I saw they had written their own DSL to do streaming analytics (Juttle). Anyone willing to do this must be crazy or brilliant, so I applied (either is quite a bit of fun to work with).
Surprisingly, I was pulled in for an interview rather quickly. My background tends to polarize employers, with the majority leaning to “so, are you going to need a leave of absence to go to divinity school next?” Jut, though, appreciated my diversity of experience and offered me a spot in the nascent marketing department as a Customer Success Engineer.
Jut’s team was a bit older than the norm for an SF startup - as an early thirty-something, I was around the median. The place was clearly a startup, but a much more restrained and adult one than some of the rhinoplastyed donkeys. Management had experience and successful exits before, and other members of the team had non-traditional backgrounds for their roles. The feel was professional but relaxed.
Boarding the Ride
Onboarding went smoothly. I was quickly buried under a ton of systems to learn, plus a language. After being used to commuting by driving 5 MPH on 101, I found BART to be actually worse. If it was cold outside, BART was hot. If it was hot, BART was cold. It was always humid, often cramped, and according to Google, trains left Millbrae only once every 30 minutes, requiring a drive to Daly City BART to get to meetings on time if I happened to be running a couple of minutes late.
But it was fun. People were uniformly helpful. While we weren’t at Facebook or Google levels of free food, there was a fridge full of diet coke, fizzy waters, fancy fruity drinks, and all sorts of other things that required money at my previous place of employment. (Extra points to the Milano cookies and Chessman cookies). Actually, this was terrible, as I love sugar, and it was everywhere. I finally had to request celery in attempt to perform cognitive-behavioral therapy on my tastebuds.
The Ride, or Facing the Woodblock
At the top of this piece, Adams noted that Short Ride begins with an insistent rhythm on the woodblock which (mostly) never varies throughout the piece. Adams likens it to a gauntlet the orchestra must cross through. I liken it to constant pressure to get the growth curve skyward.
Startups operating quickly is a theme repeated so often that people often assume this is an inherent strength of startups. I disagree. The optimal speed for a business to move is dictated by its market, not its size. While large companies often move like dinosaurs taking a spa day at the tar pits, this can be entirely appropriate for a mature market. (If you ever tasted New Coke, you can appreciate the wisdom of sleeping on some ideas before bringing them to market).
Startups generally need to be fast because they are born lacking two important things - knowledge of their market, and funding. The goal is to learn enough about their market using the funds they have so that they can build something people want to buy. Fast is necessary not so much because someone else will build the better mousetrap first but because you may go bankrupt before you figure out what that better mousetrap is.7 It’s a big game of Battleship, launching torpedoes of VC money in hopes of hitting a paying customer before the money is gone.
Enough About the Woodblock, What About the Orchestra?
The saddest part of Jut’s closure, from my perspective, was not getting a chance to work with the team that had been assembled.8 In my brief time there it was the most collaborative place I have ever worked. The Director of QA spent several hours walking me through our systems. Engineers would go over system architecture with me or help with the details of using Juttle. When I had a customer program go missing, I had at least 2 engineers, both of our DevOps guys, and our data scientist hunting down where it could have gone missing. None of this was in their job description. I wasn’t in their silo, and I was probably lengthening their work day. (For any of you reading this, my sincere thanks for all the help).
(In case you’re wondering about the product - the missing program was ultimately found using Jut. You see, the output logs of the Jut processes were themselves piped into Jut, and so even though the original program was gone, we could use Juttle to examine the logs to find the record of the last execution of the missing Juttle. It was meta enough to make the staunchest graybeard LISPer proud).
Oracle Makes Life Miserable
A lesson for working in an SF company - during Oracle Week, telecommute. Otherwise, be prepared to face BART stations with parking filled beyond capacity before the sun rises, useful streets closed, 5 trillion people everywhere, and rain. Well, I don’t know if Oracle caused the rain, but rain it did while trying to get to a customer site through traffic flowing like traffic in SF during Oracle Week, in the rain. Not recommended.
Everyone Lies, Including Google
As The Ride neared its end, I learned something important. Google’s directions can lie. Well, fine, lie is too strong a word. “Omit useful information” (i.e., a lawyerly lie). It turns out that BART line from Millbrae ran twice as frequently as Google Maps was showing, and some of those Daly City runs could have been avoided.
The Brakes are Thrown
Sunday night prior to day 11 our CEO sent a company wide email calling an all-hands the next morning at 10AM, with no info. This was taken as a Very Bad Sign. I got to work Monday morning and discovered our weekly Bug Scrub was canceled. I went back to work troubleshooting why Linux hated my external monitors.9 We had the all hands. It was announced we were shutting down, effective essentially immediately. Well, I thought we were completely shutting down at the time. It was a confusing morning. Short Ride goes full speed throughout the whole piece until the sudden end, making a fitting final parallel.
Adams described Short Ride as being inspired by a ride in his brother-in-law’s Lamborghini: “You know how it is when someone asks you to ride in a terrific sports car, and then you wish you hadn’t?” On that, I disagree with Adams. I enjoyed my 11 days at Jut (though the first 10 really were the highlight), and was glad to make the acquaintance of an awesome team. My feelings on the matter, and where to go from here, are best summed up by the dreadfully underrated British songwriter and guitarist extraordinaire, Richard Thompson, in a song also released in 1982:
Let me ride on the Wall Of Death one more time
Oh let me ride on the Wall Of Death one more time
You can waste your time on the other rides
This is the nearest to being alive
Oh let me take my chances on the Wall Of Death
— Wall of Death10
An excellent performance at the Proms conducted by Marin Alsop can be found at https://www.youtube.com/watch?v=5LoUm_r7It8. Adams’s comments can be found at http://www.earbox.com/short-ride-in-a-fast-machine/. ↩
I know Strunk and White says to add the “‘s” to names that end in s, unless they’re a historical name, but I really hate the way it looks. ↩
I started drafting this post shortly after Jut changed status, but it took several weeks to actually finish. In the interim, Abhas Gupta made some headlines with his Donkey vs. Unicorn post. This particular donkey-discussion developed independently, and was actually inspired by this t-shirt. ↩
Just to be clear, the donkey/unicorn references in this piece are not veiled references to Jut. Anything read as a criticism of Jut is being misread. ↩
For sake of space, I will sometimes refer to the event in this piece as Jut’s closure or shutting down. As is mentioned, Jut is still around, in a different smaller form, but there’s not a good single word for this transformation that I’m aware of. Furthermore, the “Jut I knew” is no more. As a wise old Jedi once said, it’s true, depending on your point of view. ↩
Of course, under such circumstances other pieces of 20th century music may come to mind - perhaps Messian’s Quartet for the End of the World. Still, it seems best to keep that in reserve for the day when the tech bubble of which we shall not speak - let’s call it Macbeth - pops, in the process releasing a deadly unicorn neurotoxin spread by Twitter and iPhone vectors, resulting in the wholesale devastation of San Francisco’s organic, fair-trade, gluten-free, Uber-delivered fresh-squeeze juice-on-demand industry, as well as the other ancillary unicorn support and grooming infrastructure.) ↩
Eric Ries discusses this idea fairly extensively in his book Lean Startup, which I read while working on this post. Incidentally, it’s probably the best business book since The Innovator’s Dilemma. ↩
OK, not strictly true. The saddest part was having to job hunt again, but besides that. ↩
Thinkpads + MST + Linux + external monitors + 1440p = fail. ↩