Entrepreneurial Design at SVA

February 6, 2012 //

This semester, Gary and I are teaching the Entrepreneurial Design class the School of Visual Arts’ Interaction Design department. The class is required of first year students in SVA’s two-year MFA program.

In many ways – all practical ways? – our class isn’t a design class. There’s some stellar design classes at SVA, but there seemed to be less awareness of entrepreneurship, or entrepreneurial-ish things, among students.

And so our fairly-explicit goal is to maximize the number of students who will start companies. We’re using stories from founders, which we call love letters, to humanize “the mythic entrepreneur,” and we’ve tried to use the projects to give experience and quick wins. The final project is literal ramen (or cereal) profitability.

Gary and I are very bullish on designers’ ability to create transformative products and services.

While our views have been informed by working at USV, we’re not assuming the “end goal” is or should be a venture-backed company. We are assuming the internet will be relevant to whatever one does.

We’ve also designed for openness: the syllabus and assignments are available online. We’ll be tumblr’ing about the classes and posting audio recordings of many guest speakers. The students will be sharing their work online as well. Please do follow along if you’re interested.

Cherry is fascinating and grocery stores belong on second floors

December 8, 2011 //

A bunch of conversations have come out of my what comes next blog post, nearly all of them unexpected. It’s been fun – and has served as a fantastic endorsement of blogging. (As if another were needed.) Two bigger themes for the evening:

1. What do you think of X
I’ve gotten a few “what do you think of X” emails, where “X” is a new web or mobile service that wants to disrupt work in one way or another. I’m trying to think through these questions in a more systematic way, but in the meanwhile, I’ve found Cherry pretty compelling.

Cherry is an iPhone app that lets you order up an immediate car wash. Tell the service where your car is parked, and Cherry will send a professional washer to clean your car. The service has mostly been derided as a sign of an oncoming apocalypse/bubble/disaster, and maybe it is one. I don’t know, but I don’t find that a very interesting question either.

I do find Cherry compelling for two reasons:

A. “Liberating” workers
Like Opez, Cherry enables car washers to become freelancers. They don’t need to work for a car-washing shop and so can retain more of the profits generated by their work.

Historically, one service provided by a car-washing shop has been finding, staying in touch with, and retaining customers. The internet’s made it relatively easier for anyone – from a solo car washer to a car-wash shop manager – to find and retain customers: Angie’s List, Yelp, Groupon, foursquare .. the list of tools goes on and on. (And will continue to go on and on.) A motivated individual can use these tools by himself or herself or can rely on a lighter-weight solution (like Cherry) for the heavy lifting.

We’ll see more of services supporting newly liberated workers.

B. Shifting car washing from capital-intensive to labor-intensive
Over the past few decades, we’ve largely replaced people with machines in the car-washing process. (We’ve also done this in many other processes in our economy, but that’s mostly irrelevant to this argument.) This is capital-labor substitution: for years, the machines were relatively cheaper than the humans, and so we got relatively more machines (capital) and relatively fewer people (labor.)

Cherry’s reversing this trend: it’s suggesting it may be more efficient to have people and squeegees in the streets than big hoses inside of stone garages.

Why are we seeing a mix shift away from capital? Perhaps because labor is relatively cheaper than it has been. In the US at least, thousands of the people who have been laid off recently have not found productive, full-time work elsewhere. There’s an argument that these workers actually had zero marginal product in their previous occupation, and so the appropriate wage rate is zero; there’s also an argument that wages are sticky – that these workers will not accept wages commensurate with their (non-zero) marginal product – and this reluctance is keeping them under- or unemployed .. Or perhaps Cherry is an aberration, and we’ll soon be back on our capital-growth path.

2. Local real estate
The reputation-based world I started to lay out, taken to its extreme, would mean that real estate is less important today than it has been in the past.

Think about it this way: brick-and-mortar stores are generally located on the first floor of buildings to maximize their access to passers- and drivers-by. It’s been a method of discovery. How did you find your grocery store, your dry cleaner, your shoe repair shop? By “stumbling across” these places in the course of going about your day? That’s offline discovery, and that’s (mostly) how the world works today.

In a wholly reputation-based world, discovery will be wholly online where reputation can be tracked. You’ll use these reputation tools/metrics/systems to “discover” the place, and then you’ll go out and find it. It seems a stretch to imagine grocery stores and fast food restaurants on buildings’ upper floors, rather than ground floors, but I’m pretty convinced that’s what a world of online discovery, rather than local discovery, looks like.

What comes next

November 28, 2011 //

Cross-posted from the Union Square Ventures blog —

The backstory here is that I attended eight demo days this fall and saw 160 companies launch. This somehow gave me the bright idea to try to figure out the world, or where the world’s going – what better way than an accelerator-program tour to get a sense for the problems about which entrepreneurs are thinking?

Turns out the entire “explain the world” thing is pretty hard – but this post is my attempt to explain a few things at least.

What comes next
Now is a great time to be an internet entrepreneur. While much of the global economy sputters, tech companies post growth numbers other industries haven’t seen in years. Their success hasn’t gone unnoticed, and the pace of tech-company creation has quickened.

There’s more than me-tooism going on. It is, for example, easier to start a tech company than ever before – it’s easier to access startup capital, procure basic infrastructure and tools at lower prices, find and cultivate mentors, and join an accelerator program.

Accelerator programs that focus on early-stage technology companies have grabbed headlines recently. The programs are designed to be crash courses in starting a technology company and bringing a product to market. They often last three intense months, ending with a structured pitch to a roomful of investors. The most famous program is Y Combinator, based in Silicon Valley, but there are other programs around the world, including Techstars, 500 Startups, DreamIt, Excelerate, and Seedcamp.

Over the past few months, I’ve seen over 160 companies come through eight different accelerator programs. It’s a skewed group, but it captures the zeitgeist of a certain segment of the tech industry – and, I think, looking at these companies is one of the best ways to get a sense for which opportunities compel internet entrepreneurs today. Here’s a look at some of what these entrepreneurs are thinking about – and where we all might be headed:

Software is developing its own component industry
Last year, John Maeda predicted technology would become a cottage industry by 2020 “with bespoke applications made by many, rather than today’s industrialized, Microsoft-esque mass production and distribution model.” I doubt we’ve seen the last billion-dollar software company; tech companies will continue to ride the economics of software, and larger companies will use strong network effects and scale to drive consolidation. However, we’ll likely see fewer companies that need to build – and fewer companies that will build – every piece of their technology themselves.

Component sourcing has already begun to revolutionize the software industry. Take a look at Amazon’s or Rackspace’s cloud hosting services, which have lowered the price and complexity of hosting software, or the success of “outsourced” tools like Google Analytics, Twilio’s voice and text messaging infrastructure, or Urban Airship’s iOS notifications.

Entrepreneurs in this summer’s accelerator programs sliced off pieces of what it takes to build and run a web or mobile application and offered those slices as services. “I built this piece of technology three times at three different companies. I built it a fourth time, and that’s the service my company offers,” a Y Combinator CEO pitched.

Developers are the target customer for 30% of the companies I saw. Consider Parse, which provides a backend to a mobile app, and MongoHQ, which hosts instances of the open-source MongoDB. While LaunchRock offers user acquisition tools, TightDB provides a database customized for big data. ReportGrid provides website analytics, while CoderBuddy helps developers create and host websites on Google’s App Engine, and Creative Brain Studios allows game developers to deploy one game on several devices and operating systems.

Using components to build larger systems doesn’t require building monotonous systems. Consider the iPhone, made of component parts sourced from Chinese megafactories but designed in California. There’s something beautiful and unique about the iPhone that no other handset manufacturer has been able to match – even the ones that source parts from the same megafactories.

Similar to physical-product designers, when software developers start with components, they can concentrate on the core problem they’re solving and will create better products in less time. It’s exciting.

Work is shifting toward a peer-to-peer model
The first two decades of the modern internet broke industries built on distribution monopolies (e.g. music, news) and facilitated coordination between the consumer and the provider without the need for a middleman (e.g. hotels, car rentals.) The same will happen for a large fraction of our work, especially in cases where the work is standardized or employers “distribute” their workers to pools of customers.

One reason to create firms is the coordination and signaling problems of situations with imperfect information and transaction costs. As technology increases information flows and decreases transaction costs, individuals can leave their old employers and strike out on their own. Their livelihoods will still depend on providing valuable services in exchange for fees, but they’ll do so as freelancers – and on their own, they’ll capture more of the value generated by their work.

Just as blogs allowed talented writers to build audiences without being affiliated with large media organizations, and as Twitter and Tumblr allowed news- and tastemakers to succeed outside of established news or media properties, new web services will allow individuals to engage with customers without needing to work for a firm.

These free agents, disaggregated and newly empowered, can promote and sustain themselves with new tools: Opez caters to service professionals, like bartenders and hairdressers, and allows them to build followings independent of their employers, while Vayable and SideTour provide marketing and transaction-processing for neighborhood tour guides. Hiptic helps graphic designers promote their work, while Zerply helps creative professionals do the same, and InterviewStreet lets programmers show off their skills.

As technology creates new free agents, it’s also changing the notion of “work” to be less time- and location-specific. This is especially true of work that can be done easily at a distance. Workers who can’t differentiate themselves using their reputation will be commoditized. This summer, web services were launched that allow you to order up a proofreader (Kibin), blogger (Contently), tutor (LearnBop), language partner (Verbling), car ride (Ridejoy), science researcher (Science Exchange), cooking instructor (Culture Kitchen), mystery shopper (SpotCheck), or transcriptionist (Mobile Works) from your browser. The Mechanical Turkification of work has begun.

Between identified, liberated individuals and the nameless, faceless drones of Mechanical Turk lies identity: does it matter who performs the task at hand? If the worker’s background, skills, or experience matter, there’s likely to be higher variance in demand for a particular person’s services, and free agents will be sought after and chosen by reputation on services built for those purposes. Less-skilled people are likely better suited for tasks for which identity doesn’t matter, and other marketplaces that don’t include a concept of reputation will provide access to a global pool of workers.

What’s really next
Looking at activity across accelerator programs, it’s not clear which “one thing” will be next. This season’s 160 “accelerated” companies, nevermind other tech companies created outside of incubators, each seek to solve different problems and should be understood individually. (Making tricky, admittedly, sweeping essays like this one.)

Yet broader shifts like the componentization of some software development and a shift from employed workers to independent agents have begun to emerge; other shifts will become more apparent with time. These, plus the overall pace of tech-company creation, give me faith we haven’t seen anything yet.

Massive thanks to thanks to Dave, Albert, Andy, Brad, Fred, Gary, John, and Patrick for reading drafts along the way.

Technology adoption according to Mary Meeker

November 27, 2011 //

One of the most interesting slides, I thought, in Mary Meeker’s Internet Trends 2011 post was the eleventh – technology adoption curves in the twentieth-ish century in the US.

A few thoughts, still scattered:

1. This chart makes it seem recessions have little to do with technology adoption in the US. Even though the technologies tracked on the chart seems incomplete – what about the 1970s? – adoption seems to keep on goin’ regardless of macro-level economic trends. Could this be because tech adoptions is “even more macro” than a macroeconomic condition?

2. Are we really to believe there was no “new” technology diffusion between 1950 and 1990? I thought this was the US’s Golden Age of Growth. (Should we include penicillin, nuclear power, or desktop computers on this chart?)

3. The fast rise of TV is even more striking when compared with the seemingly-slower rises of desktop internet and mobile internet.

4. We’re only at 80% (desktop) internet penetration after ~20 years. By contrast, radio twenty years in was probably around 82% and TV at 95%. It’s striking because it’s slow, and it’s striking because I’m myopic – I look around and see Facebook and Twitter icons and just assume everyone in the US has internet. In fact, one out of five people don’t have internet access.

5. There’s all sorts of feel-good reasons to protect public libraries in the US, but I increasingly believe their free internet access (shared desktops and wifi) might be the most important. I’d put money on shared desktops being the most heavily-used resource in any urban library; I’d put even more money on the most-accessed site being YouTube.

6. It’s been less than a decade, granted, but mobile internet penetration is a paltry ~58% – and this is when manufacturers seem to be giving away devices. The service layer still seems to matter.

7. Let’s say there are 312 million Americans today. That means there’s approximately 187 million mobile internet users and 250 million desktop internet users. Meeker’s presentation has all sorts of stats from countries beside the US – the crux, basically, is that while there’s many people online today, “we ain’t seen nothin’ yet.”

8. Half the people who don’t have access to desktop internet don’t have access to mobile internet – okay, that’s credible perhaps – but there’s an equivalent number of people who don’t have access to mobile internet but do have access to the desktop internet. How many of these people do we think are under 18? Over 65? Or more concerning: how many are between the ages of 5 and 60 (very many), and how many will need to use the internet in the workforce (very very many, I’d assume)?

9. Why do we care so much about technology adoption? Well, it’s interesting for it’s own sake, and yes, large markets imply potential profits for entrepreneurs and investors. On a broader public-policy level, there’s some evidence of persistent effects: Diego Comin, Bill Easterly, and Erick Gong tried to show that technology adoption in 1500 explains some of the wealth of nations today. It’s an admittedly tall task, but there’s much that’s intuitive (but not inevitable) about the conclusion.

10. Despite the gloomy picture I painted above, Meeker’s chart actually makes me optimistic. We’re actually quite early in the internet technology adoption cycle, and there’s likely a whole lot more to come.

Five star rating systems

November 4, 2011 //

Recently I had occasion to think about rating systems a bit; below are some thoughts on the five-star system that seems to have become common for crowdsourced ratings. I ended up focusing on Yelp, but I don’t have anything against the service and think the team’s done a tremendous amount right since 2004. They’re simply one of the better-known users of a five-star system, and so it’s easy to talk about them – and I don’t doubt the Yelp team’s thought about these issues infinitely more than I have.

So with that introduction – I’d love to get access to Yelp’s rating data, because I bet there’s some sort of rule that goes something like “given a sufficient number of ratings for a place on a 5-star scale, 90% of places will end up with 3.5 stars.” – because really, if you look at the star ratings on Yelp, it seems like everything that’s been reviewed by more than eight-ish people has 3.5 stars. It’s totally unhelpful.

I’m not sure what 3.5 stars means anyway; the difference between 3 and 4 stars may be important, but I haven’t an idea how to think through it. To verge into the cliche, the starred system is falsely precise – and not at all accurate. Maybe the Yelp team initially hoped that the Law of Large Numbers would win out and given a sufficient number of reviews a place’s “true quality” would be revealed, but either 1) that sufficient number of reviews is massive, more than they’ve been able to generate in 5+ years; 2) subsequent Yelp reviews aren’t actually independent of prior reviews; or 3) the law of large numbers falls over for some, unique reason when the rating system involves pointy yellow icons.

My suspicion is that Yelp knows something’s wrong with their rating system, but they can’t move away from stars because they’re such a core part of the service’s culture, both internally with employees and investors and externally with merchants and users. So they’ve added histograms to break out the number of 1-, 2-, 3-, 4-, and 5-star reviews a place has, which doesn’t really solve the actual problem but is, I suppose, better than nothing.

I’ve thought some about what sort of system is better, but I don’t have a great answer. I suspect if you ask a computer scientist, he or she’d likely say “just do pairwise comparisons, and you’ll figure out which venues are better!” which is true, but I don’t believe anyone actually wants to do pairwise comparisons because that’s not how the vast majority of people think. (Q: “Which is better: City Bakery or Shake Shack?” A: “What are you talking about? Better for what? When? Who?” etc.)

Instead – and admittedly I don’t have the language correct here – I’d much prefer a design that asks something like “Would you [endorse] this place?” or “would you [endorse] this place to your friends?” Endorse isn’t quite the right word, I think, because it’s loaded (in American English) with these notions of QVC salesyness that no one wants. Yet I also don’t think recommend is the right word, because I could like a place very much and have little desire to recommend it. Support, approve, sanction – none of those feel quite right either.

Other suggestions?

Starting a place taxonomy

October 31, 2011 //

“Place,” like “sharing” or “status,” is one of those concepts that’s become so broad as to become entirely meaningless. I hear it used most often alongside location-based applications and services, and it inevitably means something different each time.

So while talking about the idea of “place” the other day, I tried to break down the concept. I ended up with a fairly-reductionist taxonomy – one that, perhaps with a good deal more work, could become helpful.

Category	Example place	Example phrase(s)	Related service	Time dimension matters
Utility¹	Shake Shack	“How long is the line at Shake Shack today?”; “Who do I know at the Shake Shack right now?”	Foursquare, Sonar, LocalMind, LocalUncle, Twitter	Tremendously
Utility²	Ippudo	“Where’s the best ramen in New York City?”	Foursquare, Yelp, Quora	Less so
Wanderlust	Istanbul’s Blue Mosque	“I really want to go here (but I have yet to go)”	Pinterest, Tumblr, Flickr	Oftentimes not
Secrets	Corner of 6th b/w Bowery and 2nd	“This is where I was standing when Brad called and offered me a job at USV”	??, private geotagged Flickr photos? Path posts?	Timestamp matters more than time

This table is incomplete and sparse. But if I can draw premature conclusions, it seems we’ve designed pretty alright services for the first two categories, alright services for the third, and absolutely nothing for the final category.