Monday, February 22, 2010

Why diversity matters (the meritocracy business)

Diversity is the canary in the coal mine for meritocracy. As entrepreneurs, more than any other industry, we’re in the meritocracy business. The companies that make decisions based on merit, rather than title, politics, or hierarchy execute faster and learn faster than their competitors. For startups (and other innovators), that’s a decisive advantage.

So when a team lacks diversity, that’s a bad sign. What are the odds that the decisions that were made to create that team were really meritocratic? That’s why I care a lot about diversity: not for its own sake, but because it is a source of strength for teams that have it, and a symptom of dysfunction for those that don’t.

There’s been a lot of hand-wringing about gender equity in the high-tech and entrepreneurship worlds lately. This is a good thing. (I want to especially recognize Vivek Wadhwa and Brad Feld for their leadership.) We have a lot of introspection to do. By any objective standard, we ought to be diverse. Our industries are young, so there hasn’t been time yet to become encrusted with too many traditions that exclude outsiders. The work itself, especially in startups, depends primarily on intelligence, communication, creativity and empathy. Even the most radical Bell Curve-style thinkers have to concede that even if there are differences between men and women in the distribution of these traits on average, these curve have substantial overlap, and there should still be a lot more of them represented in high-tech startups.

For the record, I don’t think such biological differences are one-sided in favor of men. In fact, recent research suggests just the opposite. Vivek Wadhwa and his team continue their excellent work investigating the true nature of entrepreneurship. Their recent article suggests that startups led by women are actually more successful, on average, than those led by men. This doesn’t surprise me at all, and you don’t have to support a biological determinism theory to see why. If women face structural barriers to becoming entrepreneurs, then those few who are able to overcome those barriers are probably exceptional to begin with. Or, maybe it is true that women make better entrepreneurs than men. Either way, we’ll benefit if more women are welcomed into startups and other high-tech companies.

Diverse teams make better decisions than homogenous ones. I won’t recap the academic research that underlies this assertion; for that, you should read James Surowecki’s excellent Wisdom of Crowds. The most counter-intuitive part of this phenomenon, though, is that this diversity still improves outcomes even when the added opinions are less accurate than the previous consensus. The hypothesis, which makes sense to me based on the teams I’ve worked with, is that having someone with a wacky outlier-type experience causes everyone else to reexamine their own assumptions and find flaws in their own argument. So, even if you already know best, you’ll still benefit from having others on your team to challenge you.

Diversity benefits men, too. One of the most pernicious effects of groupthink is the sense of entitlement it breeds. Teams that are complacent are less likely to challenge their own assumptions, less likely to listen to feedback and, therefore, less likely to learn. This is especially important across functional lines. I’ve seen many times what happens when a single department get’s holed-up in its own space, like the terrifying “operations cave.” Outsiders are afraid to enter, let alone make a suggestion. The safety of the group becomes an impediment to dealing with reality. Problems are easily dismissed as PEBKAC ("Problem Exists Between Keyboard And Chair") rather than as opportunities for root cause analysis. Engineers are offenders in this category too, but so is any gender-segregated activity, like an all-female PR or marketing team.

Diversity is not the only requirement for making good group decisions. Two others – that each team member give their input independently and that the results be objectively aggregated – are also key parts of building a meritocracy. To be clear, though, this diversity refers only to diversity of opinion, not necessarily to demographic diversity. So why is demographic diversity important?

Demographic diversity is an indicator. It’s a reasonable inference that a group that is homogeneous in appearance was probably chosen by a biased selector. Even if men have an innate advantage at software development, the gap would have to be massive in order to explain why startup after startup has an all-male team.

Another explanation is that of compounding effects throughout the “pipeline” of candidates. In school, women are discouraged from pursuing math and science. Smart women are often stigmatized. Teachers aren’t encouraging. Women are paid less than men. And sexism is surely implicated in many career decisions. These effects compound, becoming larger over time. But this can’t possibly be the whole story. If it were, entrepreneurship would have a much larger proportion of women at every level compare to other industries – because tech entrepreneurs are younger than successful people in most other fields. There’s much less of a career ladder to climb, and therefore less time for these compounding effects to add up.

Plus, technical skills are easily learned – in my experience – by anyone with sufficient IQ and determination. I have personally taught many “non-technical” people to program – graphic designers, QA folks, even artists and animators. When those people have the chance to contribute in a meritocratic environment, they have many opportunities to learn and grow. So why aren’t more women finding themselves in the position to get started on their 10,000 hours?

With identity issues, there is no dichotomy between symbolic action and substantive action, thanks to a human tendency known as priming. In psychology research, when subjects are simply reminded of some aspect of their identity, and then measured performing a task, the pre-task activity “primes” their brain in ways that affects their performance. If the priming involves a negative stereotype, it has negative effects. So, for example, asking men and women about their gender before giving them a math test produces a significantly different result than asking other innocuous questions. (See this study among many others: http://www.nyu.edu/nyutoday/article/430)

Even the fact that a startup is all-male can make it less likely that a women would want to join. Even worse, it might even affect her performance in an interview. And just solving the gender imbalance might not be helpful, if the solution involves yet more negative stereotypes. Which is why affirmative action doesn’t help. A company with an all-female support staff sends, in some ways, an even worse signal.

Understanding these problems can make the solutions seem intractable. After all, if you already have an all-male startup, you’re already at a disadvantage when it comes to hiring the very women that could fix that imbalance. But priming cuts both ways: when we go out of our way to affirm meritocracy, it actually improves everyone’s performance. In fact, explicitly making meritocracy a value is actually better than rejecting stereotypes – by calling attention to the stereotype, you’re still engaging in priming.

Instead of focusing on programs designed to specifically benefit any one group, I think our focus should be on making our companies as meritocratic as possible. I want to start with the easiest suggestion I can think of, one that I’ve personally used with great success. I first tried it as an experiment after reading in Blink that after symphony orchestras instituted blind auditions (where conductors can’t see who is actually playing), gender equality soon followed. In the US, women’s participation went from about 5% to 50% over the course of two decades. What’s notable about this change is that it has nothing to do with gender per se, and probably also eliminated many other forms of unconscious bias.

Now, whenever I screen resumes, I ask the recruiter to black out any demographic information from the resume itself: name, age, gender, country of origin. The first time I did this experiment, I felt a strange feeling of vertigo while reading the resume. “Who is this guy?” I had a hard time forming a visual image, which made it harder to try and compare each candidate to the successful people I’d worked with in the past. It was an uncomfortable feeling, which instantly revealed just how much I’d been relying on surface qualities when screening resumes before – even when I thought I was being 100% meritocratic. And, much to my surprise (and embarrassment), the kinds of people I started phone-screening changed immediately.

And yet, when I suggest this practice to hiring managers and recruiters alike, they rarely do it. Hiring managers say, “the recruiter would never go for it” while recruiters say, “the hiring manager won’t accept it.” What I think we’re really saying is: “I don’t want to know if I am biased.” That's understandable - it's embarrassing! Even if our biases are only implicit and not consciously held, the systems we build can still contain bias. When we change a hiring policy, especially if we do it in a visible way, we reap two benefits. We actually improve our hiring process and also signal our commitment to meritocracy.

Now, I don’t know how to conduct blind interviews for startup or high-tech jobs. If I did, I would try it in a heartbeat. But even after the initial resume screen, I think there are things we can do to ensure our decision process is merit-based. I’ve covered these in some detail in two previous posts: one on Assessing fit and one on conducting a technical interview. In short, it’s essential to keep the interview focused on the substance of the work the candidate will be asked to do, and not on gotcha or brain-teaser questions. Then, in the evaluation process, it’s essential that the discussion focus only on pertinent aspects of the candidate’s performance, again obeying the three rules of group decision-making: diversity, independence, and objective aggregation.

One last suggestion, which is a technique I learned from my IMVU co-founder Will Harvey. When it’s possible, I always believe in giving a promising candidate who interviewed poorly a chance to demonstrate their skills with a real application exercise. At my last company, for programming jobs, we’d give some candidates a chance to prove themselves by writing a real working program in just a day or two (usually, to write a version of Tetris from scratch). We’d do the evaluations of that code blind – without the person in the room. In some cases, we’d dramatically revise an opinion formed during our live interview. The work product is a more realistic test, although it requires much more work on the part of the candidate.

Beyond hiring, our actual work process matters, too. I already advocate cross-functional teams as part of the lean startup methodology. Here I will only touch on one of their benefits, which is the opportunity for people to learn new skills. When teams are capacity constrained, their natural inclination is to let work pile up. This leads to a loss of quality and an increase in cycle time, both of which are negative. For more on this subject, take a look at The product manager's lament. Suffice to say that when teams are left on their own to innovate, they take on a “whatever it takes” attitude. This often leads to people discovering new skills they didn’t know they had.

By itself, that’s a pretty minor impact. But combined with a true merit-based decision making process, it becomes much bigger. If teams are rewarded for the results they achieve, and ideas can come from anywhere on the team, there’s a positive feedback loop available for people who are willing to learn: as they take on new skills, they become more successful at their job. Contrast that with a traditional siloed department. If you only get promoted for getting better at design, you’re unlikely to get positive reinforcement for picking up some programming skills (and vice-versa). And when people who started out as designers or marketers (or other supposedly "soft" disciplines) have the opportunity to learn programming, we effectively create new pathways to entrepreneurship that bypass the traditional pipeline problems. And, like everything else I've discussed, this also cuts both ways: programmers who learn design skills are much more likely to become successful entrepreneurs, too. I'll let Dave McClure explain why, if you'd like the detailed argument.

There’s a lot more we can do than just these ideas. All the usual ideas are good ones: I support inviting more women to speak at conferences, the creation of women-centric networking events like Women 2.0 and Girls in Tech, reforming the way we teach math and science in public schools, adopting more family-friendly public policy, the creation of TiE-like organizations. And, of course, the Startup Visa will be a tremendous help as well. None of these are impossible. Last year, I traveled to dozens of cities talking about lean startups and meeting people interested in entrepreneurship. Almost everywhere, I spoke to rooms full of (mostly) men. There was one notable exception: Sweden. It’s not a coincidence; rather, it’s the result of proactive public policy. (And, before we get to that old tired argument: no, I do not believe you need to become a socialist country in order to achieve those results. I haven’t seen any evidence to support that assertion, although I hear it all the time.)

I think we could all do some serious thinking about how we can be part of the solution in our own lives. I’ve been in many rooms – and at countless events – with entrepreneurs and VC’s that I think would make women uncomfortable. The sexual jokes, the crude comments, even just the usual "success theater" chest thumping – all that macho BS. Personally, I have a moral objection to that behavior. But I recognize that there are plenty of people who disagree (and I’d welcome a chance to hear why they think it’s OK; I honestly don’t get it). We still shouldn’t tolerate it. We laugh at people who think software patents are awesome, or want to give the RIAA more power, or think big companies should have a veto power over new technologies that are “too disruptive.” Those are all positions that are coherent, understandable, and anti-meritocratic. We need to recognize that supporting a homogeneous status quo is just as dumb, and just as bad for our industry.


After all, we're in the meritocracy business.


Update: Some of the commenters both here and over at Hacker News seem to be struggling with the math part of the argument, which is that even if there are biological differences in ability, they are not large enough to explain the observed outcome. I'm not much of a statistician, so I am indebted to @hypatiadotca who shared this excellent and brief presentation by Terri Oda. It is so much better than my own attempts to argue the point that I am including it here. Enjoy:


blog comments powered by Disqus