git is a DVCS: decentralized version control system. Lots of commits flying
around, no central source of truth. Bitcoin and Ethereum work the same way.
I have a lot of personal git repos, including one where I keep the files that
make this blog. I keep them on a machine in my office, with my own ssh key.
It's 100% "off the grid": physical hardware, RAID, ssh, tarsnap, Linux.
Yet, for professional development or working with other people, I don't host things
myself; I use Github. They had me at identity—key management, email verification,
all the other account-management
before I even considered their professionally-managed hosting, good UX design, and nice
social features. Sold. (Gitlab and Bitbucket are fine, too.)
One might say I'm defeating the "decentralization" of git, using something "centralized"
like Github. That's true, but it misses the point; Github does things only Github can
do. Anyone is still free to use git directly, but Github provides something else (hosting,
identity) atop the base layer of git.
It's pretty clear we're headed for the same thing in cryptoassets: a base open-souce
protocol layer, with a bunch of businesses layered on top to handle banking integration
and compliance, done with much nicer usability than raw command-line tools. Even for highly
technical stuff like git, it's surprising what a difference a great out-of-box experience
Anyone can run the open-source software directly; most people won't, and that's
 The Rails community really understands this; their getting-started documentation is
the gold standard in the world of software infrastructure.
I've worked offsite for a variety of clients for the past few years. I tell
people about this and they tend to find two aspects of it surprising.
One: I don't travel much. The mythos of the beach-dwelling remote worker is
strong. I can certainly see benefits of working in a lower-cost area, but
on balance, I like where I live. My neighborhood is an exciting place that's
getting better every day, and I like being part of that. Plus my family
and friends are here; what good is travel if they don't come with?
Second: I lease an office, preferring to work there over a spot at home.
Patrick (my brother-in-law and former coworker) calls it my "traditional east
coast going-to-work routine" (guilty as charged). I'm a big believer in
habits over goals;
the habit of getting up and going to the office—the workplace—keeps
me motivated, and helps me put it down when I leave for the day.
It's surprising how effective having a separate office can be. It doesn't even have
to be far from home; a block or two is fine. The point is that it's a different place,
away from home, with a special, designated purpose.
Asset-light business models have been in vogue for the past few years.
"Uber for X", "Airbnb for Y", etc. I guess it's "clever" to build a hotel
chain without the hotels, or a taxi company without the taxis…just modularize
and create a platform, and take a skim off the top.
I'm temperamentally skeptical of shortcuts; perhaps MoviePass's failure will show
just how far the own-nothing-but-customers model can go.
Data and software are really strong moats, but they aren't magic. Incumbents will
figure that stuff out given enough time and investment. They move slowly but they
aren't completely stupid.
Other places where I see this dynamic:
Autonomous cars/tractors. Autonomy adds a lot of value but manufacturing is pretty
hard; John Deere might figure out (or acquire) autonomy before a startup figures
out how to make tractors.
Banking. Your startup can model credit better, but banks still have the cheapest
capital, because they have the deposit base; this is why nobody's managed to really
change mortgage lending, even though it's awful.
I live in an area with a lot of development going on. A few months ago, I noticed
two stores were about to open, and wondered which would open first.
One was a bagel shop. They were moving in to a space that used to be a sandwich
place, and had to a lot of remodeling. They'd been at it months, the sign was
up, but it seemed they just couldn't get it across the finish line.
The other place entered the race about 6 months in—that's a long head start
for the bagel shop. It was a chain coffeeshop (Peet's).
Given the six month lead, the smart money was on the bagel shop. I made a one-cup-of-coffee
bet on Twitter; surely it was bagel before coffee.
I lost. Peet's has been serving coffee two months, with no end in sight for bagels.
This got me thinking about chains and nailing the basics.
Big companies didn't get big by being incompetent. I've been inside that Peet's
numerous times since it opened and every time I'm in there, it's open, the floor is
spotless, and they have what I want. It's actually impressive and I've had many
chats with the management about how well they run the place. One specific thing
they do is a written shift checklist with perhaps 50 items on it, detailing every
little thing that has to get done from cleaning the pots to taking out the trash,
and whose job that is.
I'm still waiting for the bagel shop to open.
I hear a lot of cockamanie stories explaining how big chains conquered their
markets; they usually involve some combination of abusive vendor relationships,
labor exploitation, all sorts of borderline conspiracy stuff.
When the reality is, if you just nail the basics—starting with getting
the place open—you'll do OK.
I just finished Farsighted,
a book about decision-making.
I learned a few things, particularly the value of structuring big decisions into
explicit map-predict-decide phases, and why it helps to read fiction.
But what surprised me most, was how much of this apparently "new" stuff,
is standard operating procedure in tech companies. A few examples:
Written opinions before discussion. It's too easy to get swayed by
what louder / higher-status / more charismatic people think, when
discussing a decision. The better strategy, which was drilled into me
at Microsoft (for interviews), insists that each participant write down
an independent assessment of the situation before discussing it, to maintain
fidelity to his original opinion of things.
A lot of the impact/risk analysis came from Google's autonomous vehicle
"Linear value mapping" is intuitive to anyone who's studied linear algebra,
or optimization: treat a decision as a bunch of factors, weight them, and
score the outcome based on the summed product of the weights and how well each
decision performs against each factor
The benefits of diversity in decision-making
I agree with Johnson that decision-making should be taught explicitly in schools.
Not only is it an important life skill, but it cuts across many disciplines—stats,
psychology, economics, history, literature—onramps into other "pointless" areas,
where the applications aren't initially clear.
Maybe this is why I've always found investment decision-making so captivating;
there are few things as refreshingly broad.
Potentially controversial opinion: business email should have exactly
one person on the "To:" line.
I first heard this tip at the 2011 Inbox Love
conference, I think from someone at Twilio, and it's stuck with me ever since.
If you can't send an email to only one person, it's worth asking why:
Not sure who's responsible? Fix that, that's an organizational problem.
Afraid they won't read it? Maybe you haven't communicated the item's importance.
Huge status update? Use project/status-tracking tools.
Do you regularly send to a group? Use an alias or chat system.
It's interesting how many communication pathologies are actually org problems:
responsibility gaps, unclear decision processes, messed up reporting lines,
etc.—fix those and the communication problems will vanish.
"Extremistan" and "Mediocristan" are concepts from The Black Swan,
the second book from Taleb's Incerto.
Mediocristan is the world of central tendency. In non-statistical language,
that means most things cluster toward some sort of average—the average
is in the middle, and most things aren't too far from it: human height/weights,
the number of fans in a stadium, and the daily outdoor temperature.
Extremistan is the opposite: 100 or even 1000 data points, but adding one outlier
doubles the mean. Examples: wealth, social media. I heard the average facebook
user has about 150 friends; as of today, @realDonaldTrump has 54 million facebook
It can be easy to see Silicon Valley as "judgy" compared to the Midwest, and treat
that as a negative; that was my attitude, when I arrived.
But I thought about it and realized the value of good judgment—more charitably,
being discerning—is way higher in Extremistan than Mediocristan. In Mediocristan,
you can stumble through life,
pulling balls more or less randomly out of the urn—jobs,
careers, work—and end up pretty much the same. In Exremistan, telling
the good balls apart from the bad ones is the stuff of 10x or 100x differences in outcome.
Living in Extremistan is more stressful because decisions are more consequential.
You don't get rich in Mediocristan, but life somehow feels easier—more tractable.
You do X and the result is Y.
I see the terms "contractor" and "consultant" used interchangably. I think
this confuses what the two do, and how they add value.
In my view, a consultant adds value through expertise, while a contractor
brings project management and coordination.
Consultants show up and advise the client on what to do. A consultant's stock
in trade is knowledge and experience, and how to apply them to a particular
business problem, such as workforce reduction, revenue growth, or IT change management.
Contractors are experts in project management. Teams of contractors built the
Hoover Dam, the pyramids, and every road, bridge, and dam in the world. Contractors
know how to gather and sequence large amounts of equipment and labor, to achieve
progress against a statement of work or "scope".
I have a friend who calls himself a "software consultant". He's a contractor. He
thinks calling himself a "consultant" sounds more prestigious.
But the thing is, there's really no such thing as a "software consultant". Contractors
are defined by what they do/build; consultants by the outcomes they achieve for their clients.
I think individual consultants are paid better, but overall, way more money flows
We're having a bit of a debate in my HOA about the role of the board.
One school of thought thinks we should act like the board of a large business,
especially given the association has permanent employees and about a thousand residents.
Quoting Fred Wilson:
However, the Board should not run a company. That is the role of the CEO and his/her
senior management team. The Board's job is to make sure the right team is at the helm,
not to be at the helm themselves. Boards that meddle, that get too involved, that
undermine the management team are hurting the company, not helping the company.
The opposite perspective, which seems more popular among the "deal people" coming
from real estate, is that we should act more like the state department: a bunch
of political appointees, controlling and setting policy for a permanent staff.
I see the logic of the second but think we need to be more the first. In my
view, delegating and trusting the staff is the only viable option beacuse
there's too much to do. Our meetings are already approaching two hours every
month given our translation requirements (all of our meetings are translated
serially into Cantonese), and we're all a bunch of volunteers who might not be
around after a year or two. In my view, the right course of action is installation of
a great staff capable of running things day-to-day without us.
I suspect this is a common problem in large nonprofits and would love to hear how
others have handled it.