Musings

Two years ago I read a great book describing the creation of the computer. The period between 1930 and 1960 was one of war, paranoia, and frenzied product development. At all costs, the government would spend on the necessary resources to win. It reminded me of Bill Janeway’s description of bubbles, which I want to spend a few minutes on again this morning. 

By my understanding of Janeway’s framework in “Doing Capitalism in the Innovation Economy”, downstream price insensitivity happens when speculators fall in love with a category, from tulips to gold to beanie babies. I’ll excerpt my past post* to save a click:

In his mind, a bubble is a period of significant price insensitivity around a certain type of problem, either ‘downstream’ or ‘upstream’. The upstream direction is a case like the creation of the computer, or the internet: a government is enmeshed in geopolitics that lead policymakers and politicians to believe that at all costs a certain problem must be fixed. War tends to be a good example of that. The United States had to build a better bomb. And if that meant creating a new way of crunching numbers quickly, then they would spend and deploy as many resources as necessary to ensure that outcome. The downstream direction, on the other hand, is when speculators fall in love with a space: like tulips in the Netherlands, gold at plenty of moments in time, or dot-coms in the late nineties. In these cases, private investors, with increasing (and often alarming) indifference to price, drive the value of any individual object in the ecosystem through the roof temporarily.

In a brilliant application of design-thinking, he concludes that there are opportunities for fantastic creation in these moments, as innovation happens through trial, and error, and error, and error, and error. Errors are expensive; somebody needs to underwrite them. 

I’ve been thinking about this a lot lately, as have we all. But I want to make note of something important. Many readers know this, but not all: a lot of the on-demand economy companies that are growing the fastest — the startup household names in grocery, logistics, delivery, and transportation — lose money not just on a monthly or annual basis, like a software company. A number of these companies lose money *on every transaction*. That is, they are selling a service or a product today at an economically irrational price. The faster and the bigger they grow, the more money they *lose*, not make. While I recognize that no-revenue high growth startups have a similar challenge, I find it particularly pernicious in a space where the gross margins are negative, because the revenue suggests cashflow – it’s like having a lemonade stand where you *pay* .5¢ per glass on a hot summer day, and then when your stand takes off, you are hailed for it. Let that sink in for a moment.

There are a few conclusions worth drawing from this. 

On the one hand, if the market turns, even in the slightest, and the later stage venture financing starts to require positive gross margins (not contributing margins, or even EBITDA), the question of how to reward, retain, and protect the supply side of these marketplaces, the worker classes who act as couriers, drivers, and deliverers will become painful.

On the other hand, consumer demand can be a *very* powerful driver. And even for Uber, the biggest company in this category, growth is still utterly astounding. Most people in the world still haven’t heard of it. And the magic of getting a cab delivered to you, pushing a button to have someone take your junk, or clean your apartment, or deliver you a hot meal, may be sufficiently magical as to actually change conventional wisdom at scale. And once conventional wisdom turns, then there’s no stopping those services that can continue to execute. The consumer may be willing to keep going, while the unit economics stabilize, because our behavior will be locked in. This is quite possible. It’s still incredibly early days on the demand side.

Both of these things can, and likely will, happen simultaneously. Just like it took the dot-com bubble to create Google and Amazon, perhaps it will take a massive wave of dramatic price-insensitive negative gross margin growth to create the mobile central nervous system of the future. But many shirts will indeed be lost along the way.

*Investing In Bubbles, Pt. I

After seeing StartupLJackson’s tweet today, I was inspired to look inward and take stock of how we are performing today at Collaborative Fund on diversity. By virtue of being a vocal and relatively politically active young black man, I spend a lot of my time and energy thinking about diversity, social justice, and racial equality; when it comes to my day-by-day, it is constantly humming in the back of my mind, but I shy away from getting into the numbers and actually measuring it. I admit this is, at least in some part, out of fear of what I will find. But enough of all that.

This information is available on our website with some mild forensic work, so it isn’t groundbreaking, but counting, doing so publicly, and benchmarking against peers and the industry are an important piece of the puzzle, and I badly want to contribute from all fronts.

So, on women founders. According to Crunchbase, by 2014, 18% of tech startups they measured had a woman founder. Curiously, in New York the number was 21%, making it the better than all of the Bay Area. Las Vegas, at 26%, stood as the regional leader.

Collaborative Fund currently has 28.6% female-founded portfolio companies.

All of the numbers, including ours, are woefully low. Among those companies, too, the majority of them (>60%) had a male co-founder, as well. This number is also interesting, because it means that the total ratio of male founders to female founders is even more skewed, so that is the best possible number I could get us to from the data.

–

And let’s talk about racial diversity. For the sake of simplicity, I used Black and Latino as the relevant underrepresented minority, but I acknowledge that certain Southeast Asian, Middle Eastern, and of course Indigenous Americans are terribly underrepresented, as well. Here, the industry number is a bit harder to find, but I got to 1% (!) of VC-backed companies as having a Black or Latino founder. That is a terrifying number which makes me want to jump out the window. But such it is. Beyond that, UNH’s Center for Venture Research calculated that 8.5% of startups founded in first half of 2013 had a Black or Latino founder.

Collaborative Fund currently has 8.33% Black or Latino-founded portfolio companies.

I won’t even begin to muse on this number, because it is too frustrating.

–

I present these numbers without additional context, with no particular desire to justify or defend them, because they aren’t defensible nor are they justified. But I present them in the hopes that others will, too. In the hopes that we, the VC’s, can start keeping track of these things, and holding each other accountable.

At Collaborative Fund, our mission is to invest in companies that help push the world forward, impact culture in a positive way, and reflect our love for one another. Diversity is a critical component of that, and I am motivated to do more.

So I’ll conclude with this: if you’re working on a startup that fits our mission, has a cool brand, and you want to help us improve these numbers: email me today.

Reflecting on the passing of R.H. Coase a few years ago, I noted that his theorem was being undermined from all sides. For the uninitiated, “The Nature of The Firm” was published as an essay in 1937 by Coase, then a young British student of business who wanted to understand why the American companies were producing so efficiently and profitably. He concluded that it was because they were developing mass production, consolidating resources for distribution and communication – creating the firm. Of course, the Internet and Moore’s Law have dramatically lowered costs of production, distribution, and communication. As a result, the firm is no longer the most efficient mode for many – and soon most – industries. What has replaced it is a distributed, decentralized mode of communication and exchange. 

In The Decoded Company, the authors refer to the recent history of military strategy as an analogy to this trend. In the 20th century, generals and admirals in command centers would determine military strategy, to the minutiae of where ground forces would move on a daily basis. It was the most efficient way to ensure that the information flows were optimally organized. Today, IEDs and insurgents are notoriously hard to beat (when was the last time a war was reliably won against them). The same trend applies in product creation and business strategy. Node-powered network models beat hub and spoke models, where each of these nodes is its own microfirm.

Thus, individual agency will become a primary driver of resource allocation, in a way that it never has before. This picture tweeted by Lisa Gansky tells a very elegant story of this reality. 

We can communicate globally and are all connected, but we are decentralized, for the first time ever.

This is the main reason why political systems are more fragile than ever before, where the larger the body, often the less effective they are. But it is also why intimacy is the new aspiration; why a local AirBnB or handmade bag may be a higher form of luxury than the Ritz or Prada. And it’s why transparency, sustainability, and defensible social impact are *inexorable* characteristics of a successful 21st century business. A company can be railed across the coals because of an individual’s off-color tweet. A government’s security program exposed because of a rogue disillusioned employee. A $19 billion tech company can be created and grown to massive scale on 35 employees’ efforts. Each of us has wildly powerful agency in this moment.

As Mary Meeker’s talk earlier this week demonstrated, Millennials, the biggest generation the global workforce has ever seen, cares about meaningful work and recognition. We are nodes who want to be acknowledged as independently valuable, and want to enmesh their work and their values. (Tellingly, our managers tend to think we are motivated by money while we are motivated by meaning far more.) A decentralized, collaborative economy is ultimately one which is yours, and mine. It is one that leads with sustainability, promotes healthy living, gives the underserved access to Maslow’s needs, is mission-first. We own it, and thus we must care for it. 

I can’t tell you the number of investors that I admire who have a story about how they were offered an introduction to Company X, but turned it down before taking the meeting, only to watch the company go on to be massively successful. This is the second most frustrating feeling in venture, from my experience. (The most frustrating is when you never got an intro to meet the company, because there you didn’t even have the chance.) But it’s harder to navigate than it sounds.

also, the commonly-cited 4000 tech startups per year that would like to raise money per year is clearly a major underestimate

— Sam Altman (@sama) October 28, 2014

//platform.twitter.com/widgets.js

If Sam’s comments are true (which I strongly believe), I could conceivably take 10 meetings a day, every day of the year, and still have to say no to an inbound pitch every now and then. Of course, that is extreme, but it illustrates the point. There are so many companies. And among those, it is frankly impossible to know which among them is the next culture-changer, the next inspiring brand, the next unicorn. The top of the funnel is, thus, impossibly wide. After all, new pitches is only part of my job – supporting the portfolio companies, communicating with LPs, and growing the firm are all considerably important, too.

Different VC’s address this fundamental challenge in different ways: some limit themselves to certain sectors. Others ‘pattern match’, so they look for certain ‘types’ of founders. Others still only take introductions from referrals. I can’t help but think all of these are, in their own way, problematic. This is a big part of why diverse founders continue to lag, despite the growing chorus of concern about the issue. Plus, if I look to past success as a primary indicator of future success, I am summarily excluding newcomers (demographically or otherwise), many of whom can be the most dynamic and brilliant founders. And finally, we are in the outliers business. Most frameworks are necessarily designed to exclude outliers. How can a VC optimize the top of the funnel, and the ‘first meeting’ to ensure that they can be pleasantly surprised and maximize serendipity?

And perhaps this is naïve, and maybe the best companies and founders do fit within the pattern, but surely that can’t be true. Thoughts?

Deciding whether to implement sharing tools first, or an email importer, or to add profile picture, or even profile pages, can be a fraught and confusing exercise for young internet companies. Most founders with a strong point of view about what they want their product to do have an intuition for what the universe of features should be, but the extent to which those features should be implemented, and particularly the order of operations, has stumped many an agile development team.

Here is a framework to consider: while it’s very difficult to know which among the features you want are predictive of growth or stickiness, it is certainly the case that most feature implementations follow some logarithmic growth equation.

In these cases, the marginal utility (Y axis) of a feature is very, very high at a very basic implementation. As investment in the feature continues (X axis), the marginal utility very quickly tapers off, to the point where a large investment in the improvement of a feature will only incrementally improve the user experience. Because of this, feature implementation is somewhat counterintuitive. If you have a feature that, from the moment you implement it, things immediately start working better, leaving it *underdetermined* – that is, not investing too many more resources into developing it – may actually result in it being used more organically, creatively, and ultimately in a manner that drives better growth, while your precious product and engineering resources may be better focused elsewhere.

Not every feature implementation has the similar logarithmic growth curve. Some may be much more pronounced, whereby even the smallest investment in the feature will result in dramatic improvement in user experience, but which tapers very quickly. And others may be flatter, and will seem nearly linear. But it’s been my experience, and that of many folks I’ve spoken to, that for a cash-strapped, fast-growing startup, this curve overwhelmingly tends to apply to feature development. To that end, there are features, at basic implementation, that may very well be “table stakes” for a product to be minimally viable. If you notice that your users are hacking certain features in absence of you building them, these are hints as to how to distinguish between these growth curves.

But, careful: feature bloat can lead to a product having inertia, to a team not recognizing what the growth engines are, or isolating the really effective consumer behaviors. And so when you have a long list of features in your product pipeline, be careful not to implement *all* of them, but to focus on those whose marginal utility rises extremely fast.

Ben Bernanke wrote a great series for the Brookings Institution, where he is a Senior Fellow, on why interest rates remain so damn low in so much of the world. His major points were not really a surprise, but were very clearly laid out, and included the following considerations:

– Secular stagnation
It costs less to make a a good today than it did last year, or last decade, and so the ultimate price has gone down too, so the wages associated with that good are under pressure. People are also simply consuming goods more efficiently, thanks to ever-perfecting information about supply and demand.

– Global savings glut
Government policy in emerging markets, particularly in Asia, has encouraged building cash base over the last 20 years, perhaps in a reactionary manner to the credit booms (and subsequent volatility) in some of these markets. As a result, there are full savings accounts, but not much investment.

– Term premium
The term premium is how much extra juice you get for taking a slightly longer term view, manifesting in the expected price and yield of future bonds. Future bonds aren’t expected to be any cheaper than they are today, and so holding longer term bonds today is considered lower risk by the market.

Why does this matter to you? Well, valuations are high, unicorns and so-called decacorns abound, new funds are announced on what seems like a weekly basis, and we are trying to figure out when it’s going to stop. But it’s also important to understand why: while most of the discussion has focused on how many more startups there are today, and how much bigger the internet is than the last big boom (an order of magnitude), the interest rate trend is a very interesting one to watch, and a highly relevant one. 

There is robust revenue growth in the private high-technology sector, and decent liquidity because of the cash-rich big companies: there are real yields in this business (double digits, sometimes even > 20% IRR) where it’s very hard to find them in traditional investments. As such, professional investors (individuals, institutions of all types) are looking for more upside, and are more willing to take risks on the alternative investments classes (of which venture is a part). And so at the Limited Partner level – investors who fund venture capital firms – there is eagerness to get access to yield, just like at the VC level there is interest in getting access to valuation markups and big exits. 

The points Bernanke made, by the way, are points of view about why interest rates are persistently low, but are not the only points of view. Many smart people claim that secular stagnation is indeed partly because of deflationary tech (which creates efficiencies and thus lowers prices), but also because regulation lags, and in some cases undermines, investable opportunity in parts of the world where there should be higher ‘natural’ rates of GDP growth. And others still might argue that secular stagnation is a function of the rich countries having the rug pulled out from under them in 2008, but since the world’s financial systems are so intertwined, maybe they have put low-interest overhangs on growing countries who have looked to rich countries for liquidity. And finally, some people think that any notion of ‘macroeconomic analysis’ at the global level is silly, since the factors in individual countries (government, natural resources, demographics, historical asset bases) are wildly diverse. What do you think?

I am not an economist, but I do enjoy thinking about these things, because they do affect our business.

In Brothers Karamazov, one of my favorite ever novels, Katerina Ivanovna suggests the brilliant notion of ‘second-order love’. At one point in the story, in saying “I love you”, she means “I love (that I love you), but when it comes down to it, I don’t fully love you”. We colloquially refer to this type of circumstance as someone who has been ‘idealized’. That is, I’m experiencing a second-order volition, or will, more strongly than the first-order volition.

Professor Frankfurt of Princeton uses the drug addict to illustrate the complexity (and explanatory power) of these cases. A drug dealer has a second-order volition that is “I do not want to do drugs”. But he has a first order volition that is “I want to do drugs.” In that case, the volitions compete directly, or even perfectly contradict. And most often, although the ‘higher order’ volition may be the more conscious one; the one more closely activated in thinking about my will, the ‘lower order’, more ‘wanton’ volitions are often the more effective ones. 

To this point, I was thinking over breakfast with a friend this morning about how neatly that concept applies to great founders. When you think “I care about being a founder” are you referring to a ‘higher-order’ or a ‘lower-order’ will to be a founder? I use founder as a placeholder, because this can apply to any career. But since the notion of ‘being a founder’ has so much currency, baggage, and implication tied to it, I think that’s a uniquely relevant example, where the first-order and second-order volitions are often in competition, or are contradictory.

My point of view: the best founders love the idea of being a founder, and also are passionate the problem they are going after. Scott has great framing with this tweet: 

I’ve written about this before, but I like this framing. A great founder, then, may be one who has a first-order desire to solve something AND a second-order desire to start something. What do you think?

Should calculus be required in schools? What about statistics?

I have had this debate a number of times over the past few weeks with folks from very different fields (medicine, finance, software, education). 

The pro-calculus team says:
— understanding the rate of change is critical to how quickly, or in what manner, a set changes in time, or a number grows. This is important for measuring a disease’s propagation, or understanding your 401k’s compounding interest; infinite series are what recursive functions in computer science were made for!

The anti-calculus team says:
— when was the last time I calculated a differential? I didn’t need calculus to understand the concept of something growing at a faster rate. The vast majority of medical practices don’t really need calculus or physics as much as they need social science, chemistry, and biology. It’s silly to require it. Except in sophisticated finance or quantitative economics, I never need to integrate a function. And those skills are learned on the job, or in a CFA or FINRA exams.

Let’s move to statistics: I increasingly feel strongly that it *should* be required. Harshil Parikh from Tuva Labs said: “we are going from an algebra and calculus society to a statistics and quantitative reasoning one, so our pedagogy needs to adapt.” Nate Silver’s recent rise to prominence and the Moneyball and statistics driven approaches to sports and politics represent the beginning of an age where we can measure everything. 

Understanding what it means for a data point to be statistically insignificant, or the difference between the mean and the median are critical and very widely applicable tools, certainly the way that a rate of change is. How can an 11th grader get an intuitive sense for that New York Times chart showing that 2014 was a record year in temperature? Or, better yet, for looking at a map of race and poverty distributions in a city and being able to draw real, thoughtful conclusions about the data?

Data scientist is a relatively new job — maybe like software developer was in the early 90s. And just as the entire software stack has developed to dramatically lower the barrier to entry, and make the notion of ‘coding’ a toolkit that a wider and wider segment of the population can access and effectively use, it stands to reason that manipulating data sets to draw conclusions about, say, how to better coach your soccer team, or manage your grocery shopping, or optimize your boss’s calendar, ought to — and will — follow. 

Data literacy has been described by the Secretary of Education as a national priority, but is it more important than Calculus? Are Leibniz and Newton turning over in their graves at the prospect of high school students graduating with statistics but not calculus? Should stats replace calculus?

This morning I read Matt Yglesias’s incredible post on Vox about the apparent oncoming collapse of the democratic political system. His view, as described in the article, is that political parties are increasingly ideological; that parties are using the extreme political tools of the constitution to create zero sum scenarios in policy-making, eroding the effectiveness of the democratic process.

I thought it was interesting (and Chait’s response just as much so) because it got me thinking about the future of governance more generally. And so, a quick detour. The on-demand movement of the last half decade, powered by Uber, Lyft, Instacart, Shyp and others, is here to stay. The consumer of the day can come to expect that they can access goods and services from their mobile device, and that those goods and services will be at your doorstep day-of, and even hour-of. My friend Semil suggests that this is table stakes for companies developing consumer experiences today. There are some concerns about the unit economics of these models — many lose money on each transaction — and I agree there are many growing pains, some of which will be scary, to address before we realize the on-demand economy. But where my interest is actually most piqued is not in the sustainability of today’s platforms. If the consumer behavior has indeed changed, the platforms will — or should — adapt to continue to meet the demand. As of today, the platforms provide three primary services:
— Payment processing
— A fraud protection mechanism (insurance, background check, minimum quality)
— A dispatching service that dynamically matches work requests

What’s shared by all these platforms: *drivers, using their own vehicles*: It makes sense. On-demand in this environment means ‘delivered to me’ and someone has to, well, do the delivery. BYOV (bring your own vehicle), as Jonathan Matus from Zendrive calls it.

What’s *missing*, however, from all these platforms: *agency and ownership*. The BYOV class are, for the most part, considered independent contractors (for most, but not all of these companies) and as such do not get benefits, but also don’t get equity, or any other access to any of the decisive authority in managing the platforms. And they are the most important piece of the puzzle, after all. So what if we open these platforms? 

What’s to stop a community of drivers from partnering with a payment processor, integrating off-the-shelf dispatching technology, and going to market with the democratic version of these platforms? Is that like proposing a social networks be more open, which seems to have failed time and again?

I would be interested in the platform that offered governance tools to these drivers, so they could organize, impose a collective will, and manifest that will themselves. Perhaps it would be an even more decentralized marketplace, or a razor-thin implementation of an on-demand company, with a management layer that was elected, or appointed, but more like a public administrator than a CEO. How would they vote on best practices? How would they design their amendment process? What about voting on the platform fee? What would a democratic BYOV company look like?