“It is the human friction that makes the sparks”

From Jonah Lehrer, Brainstorming Doesn’t Really Work (via Stowe Boyd):

Building 20 [a scene of incredible innovation at MIT] and brainstorming came into being at almost exactly the same time. In the sixty years since then, if the studies are right, brainstorming has achieved nothing—or, at least, less than would have been achieved by six decades’ worth of brainstormers working quietly on their own. Building 20, though, ranks as one of the most creative environments of all time, a space with an almost uncanny ability to extract the best from people. Among M.I.T. people, it was referred to as “the magical incubator.”

The fatal misconception behind brainstorming is that there is a particular script we should all follow in group interactions. The lesson of Building 20 is that when the composition of the group is right—enough people with different perspectives running into one another in unpredictable ways—the group dynamic will take care of itself. All these errant discussions add up. In fact, they may even be the most essential part of the creative process. Although such conversations will occasionally be unpleasant—not everyone is always in the mood for small talk or criticism—that doesn’t mean that they can be avoided. The most creative spaces are those which hurl us together. It is the human friction that makes the sparks.

I think this underscores one of the main reasons remote early-stage projects often fail. We mistakenly think of brainstorming as something you can do in meetings, and teaching as something you can perform through carefully composed documents or lectures.

I was part of a number of failed remote R&D attempts. The one time it worked was when we decided to abandon meetings, project documents, tracking tools, etc. Instead, we got a high quality speakerphone so everyone could overhear everyone else’s conversations, and we left it on all day, every day. It wasn’t the same as being together in person, but we did manage to get some of the human friction back.

 

And then, suddenly, it works

The other day a friend was demoing a new app he was working on. My first reaction was: “Yeah, yeah. This is nicely executed version one of those ideas I’ve seen 50 times.” My second reaction was: “But I could say that about pretty much every successful startup I’ve seen over the last 10 years.”

Most of the time, important new ideas don’t succeed on the first attempt or even the first ten attempts. But then they do, and it seems to happen suddenly. It’s hard to tell why this is. It’s probably a combination of timing (riding some fundamental shift in technology or culture), and execution (getting the product just right).

An idea getting tried over and over tends to be a positive signal (which is one reason that competition is overrated). It’s very easy when you spend lots of time around startups to get cynical. You could tweet and blog predictions that every new startup will fail and how the ideas are derivative and you’d be right 95% of the time. The hard part – and what matters for founders and investors – is figuring out the right mix of timing and execution to finally get it right.

Recruiting programmers to your startup

Here are some things I’ve learned over the years about recruiting programmers* to startups. This is a big topic: many of the points I make briefly here could warrant their own blog posts, and I’m sure I’ve omitted a lot.

– The most important thing to understand is what motivates programmers. This is where having been a programmer yourself can be very helpful. In my experience programmers care about 1) working on interesting technical problems, 2) working with other talented people, 3) working in a friendly, creative environment, 4) working on software that ends up getting used by lots of people. Like everyone, compensation matters, but for programmers it is often a “threshold variable”. They want enough to not have to spend time worrying about money, but once an offer passes their minimum compensation threshold they’ll decide based on other factors.

– Software development is a creative activity and needs to be treated as such. Sometimes a programmer can have an idea on, say, the subway that can save weeks of work or add some great new functionality. Business people who don’t understand this make the mistake of emphasizing mechanistic metrics like the number of hours in the office and the number of bugs fixed per week. This is demoralizing and counterproductive. Of course if you are running a company you need to have deadlines, but you can do so while also being very flexible about how people reach them.

It is sometimes helpful to think of recruiting as 3 phases: finding candidates, screening candidates, and convincing candidates to join you.

– Finding means making contact with good candidates. There are no shortcuts here. You need to show up to schools, hackathons, meetups – wherever great programmers hang out. If your existing employees love their jobs they will refer friends. Try to generate inbound contacts by creating buzz around your company. If you have trouble doing that (it’s hard), try simple things like blogging about topics that are interesting to programmers.

– Screening. Great programmers love to program and will have created lots of software that wasn’t for their jobs or school homework. Have candidates meet and (bidirectionally) interview everyone they’ll potentially be working with. If the candidate has enough free time try to do a trial project. There are also more procedural things that can be useful like code tests (although they need to be done in a respectful way and they are more about getting to know how each side thinks than actually testing whether the candidate knows how to program (hopefully you know that by this stage)).

– Convincing them to join you. This is the hardest part. Great programmers have tons of options, including cofounding their own company. The top thing you need to do is convince them what you hopefully already believe (and have been pitching investors, press etc): that your company is doing something important and impactful. The next thing you need to do is convince them that your company is one that values and takes care of employees. The best way to do this is to have a track record of treating people well and offer those past employees as references.

A few things not to do: you will never beat, say, Google on perks or job security so don’t even bother to pitch those. You’ll never beat Wall Street banks or rich big companies on cash salary so don’t pitch that either. You’ll never beat cofounding a company on the equity grant, but you can make a good case that, with the right equity grant, the risk/reward trade off of less equity with you is worth it.

Finally, I’ve long believed that early-stage, funded startups systematically under-grant equity to employees. Programmers shouldn’t have to choose between owning a fraction of a percent of an early-stage funded company and owning 50% of an unfunded company they’ve cofounded. Naval Ravikant recently wrote a great post about this:

Post-traction companies can use the old numbers – you can’t. Your first two engineers? They’re just late founders. Treat them as such. Expect as much.

Making those first engineers “late cofounders” will dramatically increase your chances of recruiting great people. This is a necessary (but not sufficient) condition for getting the recruiting flywheel spinning where great people beget more great people.

* As someone who personally programmed for 20 years including about 10 years professionally, I preferred to call myself a “programmer.” Some people prefer other words like “hacker” “developer”, “engineer” etc. I think the difference is just uninteresting nomenclature but others seem to disagree.

What jobs are users hiring your product to perform?

One of Clay Christensen’s favorite concepts is that instead of dividing your customers into segments and asking which features each segment would like, you should think about what “job” the customers are “hiring” you product to perform. Here is an example:

A fast-food restaurant chain wanted to improve its milkshake sales. The company started by segmenting its market both by product (milkshakes) and by demographics (a marketer’s profile of a typical milkshake drinker). Next, the marketing department asked people who fit the demographic to list the characteristics of an ideal milkshake (thick, thin, chunky, smooth, fruity, chocolaty, etc.). The would-be customers answered as honestly as they could, and the company responded to the feedback. But alas, milkshake sales did not improve.

The company then enlisted the help of one of Christensen’s fellow researchers, who approached the situation by trying to deduce the “job” that customers were “hiring” a milkshake to do. First, he spent a full day in one of the chain’s restaurants, carefully documenting who was buying milkshakes, when they bought them, and whether they drank them on the premises. He discovered that 40 percent of the milkshakes were purchased first thing in the morning, by commuters who ordered them to go.

The next morning, he returned to the restaurant and interviewed customers who left with milkshake in hand, asking them what job they had hired the milkshake to do. “Most of them, it turned out, bought [the milkshake] to do a similar job,” he writes. “They faced a long, boring commute and needed something to keep that extra hand busy and to make the commute more interesting. They weren’t yet hungry, but knew that they’d be hungry by 10 a.m.; they wanted to consume something now that would stave off hunger until noon. And they faced constraints: They were in a hurry, they were wearing work clothes, and they had (at most) one free hand.”

The milkshake was hired in lieu of a bagel or doughnut because it was relatively tidy and appetite-quenching, and because trying to suck a thick liquid through a thin straw gave customers something to do with their boring commute. Understanding the job to be done, the company could then respond by creating a morning milkshake that was even thicker (to last through a long commute) and more interesting (with chunks of fruit) than its predecessor. The chain could also respond to a separate job that customers needed milkshakes to do: serve as a special treat for young children—without making the parents wait a half hour as the children tried to work the milkshake through a straw. In that case, a different, thinner milkshake was in order.

There are at least three obvious ways to apply this concept: 1) when searching for startup ideas, think about jobs people want done that they can’t currently get done, 2) when thinking about how to fix or improve your product, understand why existing users are hiring your product (or should be hiring your product) and try to improve those experiences, 3) when analyzing markets, segment companies by the jobs they are hired for. Sometimes products that might appear similar (e.g. two photo sharing apps) are actually hired for very different purposes, and are therefore misclassified as competitors.

Making industries “garage ready” for startups

One of the most important events in the history of modern computing was the advent of “fabless” (“fabrication-less”) semiconductor companies.  The story of fabless semis is similar to the recent history of internet startups: various forces led to an order-of-magnitude reduction of startup costs, which then led to a surge of innovation.

Before the 1980s, if you wanted to invent a new semiconductor, you had to both design and manufacture it. This meant you had to build a large manufacturing plant, something only large companies like Intel, Motorola, and IBM could afford. Hence, semiconductor design was generally too expensive for venture-backed startups.

In the 1979, two computer scientists published a seminal book that argued for the separation semiconductor design and manufacturing. Followed by years of investment by DARPA and others, an industry emerged where chip designers used software (“EDA software”) to design and test semiconductors, and then sent standardized specifications to “foundries” that did the manufacturing (most of which were located in Taiwan – the largest in the world to this day is Taiwan Semiconductor Manufacturing Company).

This dramatically lowered the cost of starting semiconductor design shops, and in turn led to a massive wave of startup innovation. These startups designed chips for cell phones (Qualcomm), Wifi (Atheros), computer graphics (Nvidia), and much more.  Most were funded by venture capitalists and located in Silicon Valley.

Tech sectors tend to get really creative when they become “garage ready”:  a Steve Jobs and Steve Wozniak, or a Larry Page and Sergey Brin, can, with very little capital, change the world. It happened with semis in the 80s and happened in the 90s and 2000s for internet companies.

Eventually every vertically integrated, capital-intensive sector becomes garage ready. Someday, for example, we will have “fabless” gadget design and biotech research, enabling a small shop in Brooklyn or SoMa to create an iPhone killer or next-generation cancer drug.