How to Think About Coworking

Are WeWork, Industrious, Knotel and other coworking startups overvalued?

I think they probably are and I’ll explain why, but in some ways that’s the least interesting question you can ask about them.

Whether or not these companies ever grow into their current valuations, there are hundreds of thousands of people working in their offices, and they’re influencing the standard office environment for millions more. Many of these influences will persist whatever happens to the current round of startups, including second order effects that no one’s discussing at all.

1. What are they really offering landlords?

In a sense, urban office owners were the largest initial “investor” in coworking startups, in the form of below-market lease terms. But even if you believe the startups will eventually melt down, it doesn’t follow that these landlords were all suckers, because they weren’t making the same bet as an equity investor.

In many locations they’re filling space that would be hard to lease to a standard corporate tenant — mid-block, low floors, older buildings — and in others they’re looking for a “halo effect” that will help lease the rest of the property.

When you get out of the city to some more recent entrants, there are further considerations. A mall might subsidize coworking just to get more daytime foot traffic. A suburban office park may see it as an amenity for their other tenants, or an incubator for growing companies they can recruit into standard leases.

Another thing to notice is that as money pours into the startups, they’re starting to accept higher rents, guarantee leases for a bit longer, and cover more of the build-out costs. Even if you took the cynical view that coworking was just a straight transfer of value from “old economy” landlords to tech investors, you’ve got to admit that at certain locations that subsidy may now be running in the other direction.

2. Why can’t someone else do it?

Whatever the direct rationale for landlords, there’s also a research element: they’re trying to understand what these startups are doing differently from Regus and all the other similar concepts that they’ve seen in the past. And while those differences are very real, none of them are proprietary.

So even if the bulls are right about continued rapid growth of the flex/coworking market, that doesn’t mean the economics will hold up or anyone will maintain their current market share. Because whatever the network effects of being a “member” of a coworking brand, they are not strong enough to be a meaningful barrier to entry.

Now, as the model expands to higher profile locations, larger corporate tenants and more revenue-sharing arrangements rather than standard leases, you can see a different risk emerging for traditional landlords and brokerages — not “disruption” or displacement, but disintermediation.

The best example of disintermediation in commercial real estate is the hotel sector. First the brands and then the online travel agents have captured staggering amounts of value by positioning themselves in between owners and guests in ways that increase their leverage with both sides.

If you squint hard enough, you can see a few of the same dynamics in the office business, especially as coworking share grows and leases get shorter.

But they’ll never get anywhere near the nightly “leases” of hotels, they don’t have the same basic trust problem that hotel brands solved, and even the strongest proponents of the “office as hospitality” trend are not suggesting anything as management-intensive as an actual hotel.

So while the startups have jumpstarted the growth of shared and short-term office space, in the long run the operation and leasing of that space is likely to be as competitive as ever, and it’s not surprising that so many real estate companies have already started their own coworking platforms.

There’s room for lots of different operators to succeed, and like the hotel brands and travel websites they will certainly consolidate, but it’s hard to see how anyone corners the market and extracts the kind of sustained excess returns that VCs are looking for.

3. What are they really offering tenants?

This one’s easier to answer. They’re offering an immediate move-in, a shorter lease term and most importantly, greater density: the square footage per worker in a coworking layout can easily be half of what it is in a standard office. And while some of that is shared facilities, most of it is just less space to work in.

How about collaboration, innovation, or the motivational decor?

It’s hard to say. On the one hand, as I wrote above, the differences are very real between new coworking space and older temp offices, which felt like the DMV. It’s no good saying “this is just Regus with a different name.” Whatever you think of the slogans on the walls, these are much more appealing work environments.

On the other hand, there’s really no evidence that density, open floor plans, and floating desk arrangements make employees more productive, and there’s plenty of evidence that they have the exact opposite effect. I wrote about this in more detail last year, and some prominent voices in tech have made related arguments about coworking, like Y Combinator’s Sam Altman:

“Hey, startups, we’ve got this really cool coworking space. There’s free espresso and colorful couches. Come.” You get the people you deserve when you do that…

The average level of ambition and willingness to work hard at a coworking space is incredibly low. There’s this reversion to the mean that is not what you want in your life.

So when you get past the lease terms to some of these fuzzier benefits, it’s hard to disentangle what companies say they’re getting from coworking, what they really think they’re getting, and what they’re actually getting.

4. What happens in a downturn?

If you believe that individual coworking demand will be “counter-cyclical” because laid off “creatives” decide to pay $500/month out of their own pockets to ride out the next recession in style as freelancers, then I have a bridge to sell you. If you think they’ll all found startups and raise other people’s money to pay for coworking memberships at a faster rate than today, that’s not much more plausible.

Now, if you want to argue that corporate tenants in a recession will become even more inclined towards shorter lease terms and greater density, that’s getting more reasonable. But remember that there are now lots of large corporate tenants already in coworking, and often that already is their flex space. If you’re IBM (or whoever) with most of your employees in a given market located in your permanent office with eight years left on the lease, and a few more in a coworking office with just six months left, and you start reducing headcount, guess where you’re vacating first?

And even if net demand somehow increases, there’s also more vacancy in the market competing with you, much of which can be reconfigured as coworking if that’s where the demand is.

So it’s hard to avoid the conclusion that one way or another, the spread of coworking will make the next downturn deeper for the entire office market, even for some of the landlords who never touched it. Long leases are a shock absorber for real estate cycles, and coworking is making the average effective lease term a little shorter.

What about the famous asset/liability mismatch? Will some landlords cave when the coworking operator demands a rent cut or a better revenue split?

Absolutely, but not all of them. And if that leverage with landlords is part of your bull case for current valuations, you’re still not thinking about the whole picture. Companies valued at 20x revenue need to grow revenue, not cut expenses. The scenario where they even have time to haggle with their current landlords is one where their valuations have collapsed; it may be a way for their late stage investors to recover 10% of their money instead of 5%, but it’s not a way to go from 2x returns to 3x.

5. What are the second order effects?

I have a theory about this that I haven’t seen anywhere else. I think coworking is a Trojan horse for working from home.

Remote work has been a recurring bogeyman for office real estate since the term “telecommuting” was coined almost fifty years ago. Many large companies have experimented with remote work policies and changed their minds. But maybe it’s a bank shot that needed to happen indirectly.

Many companies have slowly made their offices as cramped, noisy and distracting as possible, all in the guise of “collaboration” and “community.” Whether they believed the hype or they were just spinning the cost reductions, they’re probably not going back to lower-density layouts and more space.

But employees can solve the problem from the other direction, and I think we’re already seeing that. Everyone I know who works in a coworking space treats it as less of a full-time “home base” than they would a standard office, and spends less actual time there. 

This seems to be especially true of the technical talent at startups. If you have a full day of coding ahead of you, why would you go even a few subway stops to do it in a tiny glass-walled office in a coworking space, crammed in with your non-technical colleagues who are on the phone all day, when you could open up the same MacBook Pro at home (or a coffee shop next door) and do the same work with fewer distractions?

You can see how this evolving hybrid model is where more companies will end up, with one or two “in the office” days per week.  And maybe what coworking is really giving them is a way to accelerate this transition, by selling them even more density and calling it something else.

Satellite offices that reduce commute time are the first stage of remote work (this is actually what “telecommuting” originally meant) but the harder it is to be productive in those offices, the faster we’ll pass through that stage. And in a downturn when you’re trying to retain key employees without paying them more, and your marginal dollar of rent is paid on a short-term per-head basis, letting them go fully remote is an easy win-win decision.

If you don’t own any office buildings, is this even a problem? In contrast to open plan layouts, much of the research on working from home shows real benefits in productivity and well-being.

And we may have reached a tipping point in terms of technology. Here’s Stripe just last month on their new remote engineering “hub”:

…the technological substrate of collaboration has gotten shockingly good over the last decade… Google Docs, Slack, git, Zoom, and the like deliver high-bandwidth synchronous collaboration on creative work. The experience of using them is so remarkably good that we only notice it when something is broken…

While we did not initially plan to make hiring remotes a huge part of our engineering efforts, our remote employees have outperformed all expectations…

I could be wrong! It’s just a theory. But the one forecast that’s guaranteed to be wrong is the one I keep seeing everywhere else, which is hockey stick growth in coworking/flex market share without any offsetting changes in work patterns.

In fact, the more you believe in coworking — the more that it’s really something new and different from previous temp office models — the more you should be thinking about these kinds of feedback dynamics and second order effects.

What’s Really Happening to Retail?

It’s true that the “retail apocalypse” narrative is oversimplified, but the counter-narratives are not much better. They tend heavily toward the anecdotal (“look at how this mall replaced their empty Sears”), conspiratorial (“Amazon investors won’t put up with those low margins forever”) or just tautological (“retailers who can evolve and adapt will be survivors”).

Many of the real answers are very location-specific or retailer-specific. But at a high level, can we say anything more substantive than “Amazon and millennials are killing the mall” or “America is over-retailed” without falling into these traps?

I think we can. Here are five other angles to consider.

1. Format & Location

Maybe you know the story about suburban enclosed malls killing downtown retail, then being targeted themselves by the rise of open-air formats like power centers, outlet centers and lifestyle centers — many of which are also now struggling.

Can you really blame Amazon for this? Much of it is just old-fashioned on-the-ground competition, as shoppers move on to new master-planned town centers, open-air malls and mixed-use projects that are more pedestrian-friendly and often closer to home.

(This is one problem with turning a dying mall into a “town center” by replacing failed tenants one at a time: if the surrounding demographics support something like that, there’s a good chance someone has already done it.)

Landlords like to point out that replacing defunct tenants has always been part of their business. But the real estate has its own cycle of obsolescence, even if it’s a longer one.

2. Winners & Losers

The key difference between retail and other types of commercial real estate is that you’re selling cotenancy and traffic, not just space. You probably don’t care who else is in your office building, but every retailer knows exactly who they want as neighbors.

This is why a single mall can make the developer a billionaire: success in retail compounds much faster than it does for other real estate. Better malls attract better stores, which attract more shoppers, which attract even better stores, which pay higher rents, which the landlord can use to expand, which attracts more shoppers, and so on.

But this is a high wire act that depends on low vacancy and a not-too-lopsided distribution of sales, so that tenants feel they’re each getting at least as much benefit from the center as they’re giving back. And even at a mall where total sales and traffic are not declining, a larger share going to the hottest tenants (like Apple or Tesla) can start that feedback loop working in the other direction.

3. Less Specialization

Another important dynamic is the ongoing push and pull of different retail categories being combined, separated and combined again. Remember that department stores were the original malls, killing all the mom-and-pop stores downtown. Then they were lured out to the suburbs, where they struggled for decades while the smaller specialty shops in the mall (and then the category killers across the street) thrived.

Now the broader trend may be turning back towards consolidation, as the surviving general merchandisers with larger stores and lower rents find themselves in a position to cannibalize the wounded specialty stores in weaker categories (like toys) and to pile on in stronger ones.

Some of the traditional department stores are too far gone, and too compromised by their locations in obsolete malls, to take full advantage of this. But for off-mall retailers like Walmart, Target or Kohl’s, there’s never been a better environment to take share and concentrate more sales in less space.

4. Buying vs. Selling

The idea that retailers need to “adapt” and “evolve” to keep up with tech trends is actually a very dangerous one, and I wrote last month about the ways that many are “evolving” themselves to an even faster death.

It’s not that these initiatives don’t make sense on their own terms; many of them do. It’s that they tend to be focused on distribution and marketing, which are downstream from product and branding. And usually it’s the products or the brand (or both) where the real problem lies.

You can curate the perfect channel mix, recruit the best Instagram influencers, and get every customer to download your app and join your loyalty program. You can profile them to send each one exactly the right “personalized” email with the right discount code at the right time, and then ship their order from just the right store or warehouse based on real-time algorithms that maximize your profitability. Then you can spam them with customer-satisfaction surveys and feed that data back into the top of the funnel. But if they’re not excited about what you’re selling, none of it will matter in the long run. You can squeeze a little more out of your existing customers this way, but your costs to retain them (much less acquire new ones) will keep rising until you’re out of business.

This is an awkward truth that often seems to get lost on the way up to senior management. Retail has always been a combination of figuring out what customers already want and selling them something new — and the first is both less risky in the short term and more amenable to metric-driven research and management. But you have to do both.

In the long run, this is where the real advantage for physical retail will come into play. However brilliant your Instagram ads are, they’re still being delivered through a three-inch rectangle crammed with other distractions. When you get someone into a store, at least you’ve got a fighting chance to get them to look up from their phones and really differentiate your brand.

Another way to say this is that online retail is optimized for buying, but physical stores are ideal for selling — and much of our discretionary spending is on products that are more sold than bought. 

And yet many retailers have decided the “store of the future” is one that gets you to look back down at your phone instead as soon as you walk in the door — to tweet out their hashtag, take selfies with their Instagram-bait decor, scan QR codes, pull up digital coupons, and so on. They’re turning their biggest strength into another weakness.

5. Commodification and Brand Lifespan

The “lifestyle” startup brands are certainly not shy about selling, but they’re often doing it with direct response tactics that are undermining their brand value even as they build it. And digital commerce is a particularly awkward fit for fashion, where the hottest products and brands are often the least convenient or micro-targeted. So much of what makes something cool is that it’s hard to get in ways apart from price: only sold in limited quantities, in certain places or at certain times.

As with the buying vs. selling dynamic above, this tension between “convenience” and exclusivity is a slightly uncomfortable truth about human nature that everyone in advertising knows, but much of the startup world hasn’t grappled with and many legacy retailers have forgotten. By definition, if something is available online to anyone with a credit card, it’s not “exclusive.” Obviously you can still have limited editions, flash sales and “drops” online, but beyond a certain scale these are not great for your brand either, and you can target them more precisely offline.

Consumer psychology is also changing. The “Feiler Faster Thesis,” from the TV journalist Bruce Feiler, argues that the accelerating news cycle is not just a “push” from technology, but a “pull” from how much faster we all process new information and want to move on to the next thing. And similarly, shoppers are increasingly likely to abandon a particular retailer or brand, online or offline, even if they haven’t done anything wrong — just because that’s what shoppers do now, that’s what shopping is.

What all that means is that apart from a few big winners, we’re likely heading for a future with a greater number of smaller, shorter-lived brands — and most of the industry is simply not capitalized or structured for that.


Let’s recap those five points. Even if you held e-commerce penetration constant, you’d still be seeing rising store closures from:

  1. increasingly obsolete formats and locations
  2. concentration of specialty sales in fewer brands
  3. large-format / general merchandise retailers taking share
  4. self-inflicted wounds, as legacy retailers scramble for online growth
  5. a shrinking lifespan for retail and consumer brands in general

Note that they all reinforce each other in various ways — for example, the winner-take-all dynamic (#2) makes the traditional mall format even more obsolete (#1). And of course they’re all exacerbated by e-commerce growth, and they’re all feeding back into it.

The good news is that these current trends may not tell us much about the long-term equilibrium for online vs. offline sales, and in many categories the logic for physical stores is stronger than you might think.

The bad news is that with all these factors moving in the same direction, it’s likely to get a lot worse before it gets better.

Remember that George Carlin routine about environmentalism? Don’t worry about “saving the planet,” he said — the planet will be fine, it’s the people who are screwed.

Retail will be fine. But a lot of retailers are still in serious trouble, and so is a lot of retail real estate. You could unplug the whole internet tomorrow and that would still be true.

The Omnichannel Trap

I’ve been reading hundreds of retailer earnings transcripts over the last year, and I’m noticing some recurring feedback loops and circular logic in their current business strategies.

Here’s a simple and relatively benign example, not specific to retailers, that you may recognize:

  1. You promote your loyalty program (or app, or store card) with benefits that make it a no brainer for frequent or high volume customers to sign up
  2. Now you say “we need to focus even more resources on our Elite loyalty members because they drive 60% of our sales / spend 30% more on average than non-Elites / visit 40% more often…”

Of course they do, but which way does that causation run? Is it that your loyalty members are your most valuable customers, or that you forced your most valuable customers to become loyalty members? If the goal was just to get more data on them, how are you using that data to drive your bottom line? Are they spending more than they already did (not just more than other members) or are you retaining them at higher rates, or what?

I always assume there’s a lot more of that number-crunching going on behind the scenes, but after a while you do start to wonder. If they have meaningful stats, why are they giving out these meaningless ones instead? Loyalty programs have been around for centuries, but as they get more complex, data driven and expensive, does anyone really know how well they’re still working?

In any case, here’s another feedback loop that’s a little more complicated:

  1. You redirect capital away from your stores to build out your digital storefront, leading to
  2. rapid online sales growth on a low base, while the stores stagnate;
  3. the online channel is now your main growth driver, so you push it even more and start closing more underperforming stores…
  4. but then that online growth slows down sooner than expected, at a point where it’s still only 10-30% of your total sales and still a drag on margins…
  5. so you turn back to your leaner, omnichannel-enabled store fleet, where you’ve finally cleared out inventory and become “less promotional”… 
  6. but traffic is still dropping there too, and there’s nothing in your “omnichannel” toolkit to actually get more people in the door without siphoning them from your already-slowing online business.

You’ve launched this whole new channel and cleaned up the old one, you’ve done everything you’re supposed to, but your total sales are still falling. What happened?

The answer may not be so different from the loyalty program example above: you’ve been pushing your existing customers around without acquiring enough new ones. That early online growth was mostly from them, and even the first cohort of new customers that you “acquired” online were probably those who already knew something about your brand and were positively disposed to it. It only gets harder from there.

Meanwhile, your relationship with all these customers has gotten a lot more intrusive. Rather than just mailing catalogs and sale fliers, you’re flooding them with emails and nagging them at every opportunity to do more than just buy — you need them to download your app, follow you on social media, review what they bought, rate someone else’s review…

And for what? None of this is cheap for you either. What is it you actually want them to do? First it was to switch from the store to the website, then the app, then maybe you closed their store, now you’re raising their free shipping minimum to nudge them to pick up their online order in a store that’s further away… you’re “optimizing” their behavior to fit your shifting cost structure, but it’s probably driving some of them away, and in any case it’s not really growing your overall business.

Where did you go wrong? Your landlords might say: “well, you can’t make money online-only after shipping and returns, and customers also want a nice ‘experiential’ store where they can touch and feel the merchandise.” They’ve got a few of their own “omnichannel” stats that also make you wonder about the direction of causality: “Did you know that online sales drop 20% in every zip code where you close a store? That the average customer who returns an online purchase in store spends another 30%?”

And your digital marketing agency might answer “look, your stores are in dying malls anyway, your competitors are all online, they’re all doing this stuff. It may not be easy but you have to go where your customers are, there’s no turning back the clock. We just need to find the right Instagram influencer this time.”

They could both be right! For example, you may have conflated a format problem with a channel problem, and you’d have been better off moving those mall stores to more urban and open-air settings rather than closing them outright.

But let’s set that debate aside. You’ve heard it all before anyway, and what’s done is done. What do you do now?

This is where it gets interesting for landlords and creditors, and for landlords especially. Because a retailer in this position can start to look at their remaining leases and even their namesake brand as a wasting asset, where they can harvest customer data and dwindling traffic to build new brands — either online only, or with an entirely new store base.

In some cases it’s still better to give them a rent cut to stay open and avoid more vacancy even when you know they’re closing at the end of the lease — but where landlords have a choice, it’s increasingly important to distinguish between tenants in this “trap” and those who can really still turn around their core business.

The Myth of the Open Plan Office

Is there anyone who still believes that open plan office layouts increase productivity? Article after article, study after study has made it clear that the massive distraction they create outweighs any benefits from “collaboration” or “spontaneous interaction.” It’s almost a running joke at this point in the tech world:

And yet some of the smartest companies in the world keep building them, over the objections of their own employees. What’s going on?

The obvious answer is cost: office space is expensive, so companies are always trying to reduce the space per employee, and the collaboration/creativity/community stuff is just spin. But even that story only takes you so far, because at some point the lower productivity hurts the company as much as (or more than!) the employee. It’s crazy for a big tech company to save $10k/year on rent by seating a $300k/year programmer on a trading desk where they need to wear headphones all day.

So on some level, there must be people at the top who still believe the myth. In his recent book Deep Work, Cal Newport dissects two of the post-war stories of highly productive collaboration that have driven it: Building 20 at MIT, and Bell Labs in New Jersey.

Neither building offered anything resembling a modern open office plan. They were instead constructed using the standard layout of private offices connected to shared hallways. Their creative mojo had more to do with the fact that these offices shared a small number of long connecting spaces — forcing researchers to interact whenever they needed to travel from one location to another.

Originally, open plan offices were about surveillance as much as cost:

The first modern office was the Larkin Administration Building, which opened in 1906 in New York. Designed by Frank Lloyd Wright, it was based on an open-plan factory, with a giant atrium and very few walls.

“He was interested in creating big cathedral spaces for aesthetic reasons,” says Jeremy Myerson, professor of design at the Royal College of Art.

“But actually it was very useful for managers because no walls meant they could supervise workers in this open space and keep an eye on them. It was all about control.”

Every time I’m in a WeWork or similar coworking office, they seem to be getting more and more subdivided: they’re discovering that members will almost always pay a premium for a tiny glass-walled interior closet compared to a seat in an open bullpen. Which shows you that the closer the decision making is to the actual occupants, the higher the premium becomes on privacy. And if you visit even the newest and best-amenitized corporate open plan offices, you’ll find huge waiting lists for the conference rooms, people running in and out to take phone calls, and those giant headphones everywhere. (A friend at one of these told me their colleagues were starting to replace their Bose noise-cancelling headphones with the industrial ones worn by lumberjacks.)

So many of the changes in the typical office over the last few decades are about trading easy-to-measure cost reductions for diffuse, hard-to-measure costs in time and attention; think of the way that software has eliminated so much admin staffing by making everyone into their own assistant, spending hours logging expenses, contacts and meetings on a dozen different websites. Each of these trade-offs individually is so small and hard to measure (sometimes there are real efficiency gains involved) that you can see why they keep happening, even if they’re collectively a mistake. But this one is just too big to be explained that way.

This is especially true at those big tech companies with more cash than they know what to do with. In these cases an open plan office may be a kind of costly signaling device (like a peacock’s tail) or a way to keep employees distracted from the fine print on their stock options. But more likely it’s just another irrational management trend in an industry that prides itself on rationality. Hopefully we’re finally getting close to a tipping point.

The Problem with Prediction Markets

Prediction markets — websites where you can bet on politics — are rising in prominence. This election season, their “odds” are often quoted in news stories alongside poll averages.

Of course, these odds aren’t worth much if there aren’t many people betting, and there aren’t. This article by Philip Wallach is an excellent survey of the landscape, and as he points out, the US election has 50 times more betting volume at Betfair (a UK sports book) than at PredictIt, the largest current prediction market.

Social scientists have been infatuated with the idea of prediction markets for decades, but gambling is first and foremost a habit, and the problem with these markets is that there’s not enough to bet on to get people in the habit of coming back. The political world simply doesn’t offer frequent enough events that enough people care about.

These two variables — frequency and interest — can explain the popularity of sports betting and day trading, which are the two most widespread forms of gambling that prediction markets could emulate. You probably care a lot about how your favorite team or fantasy players perform, and they play often; meanwhile you may not care as much where the price of gold or Apple stock is in an hour, but the feedback is even more frequent (and traders are highly vulnerable to conspiracy theories about market manipulation, because they want to care).

Meanwhile, you may care who wins the election, but if it’s weeks or months away, what’s going to keep you coming back to the site? Suppose you came to PredictIt in June convinced that Clinton would win, saw that she was “trading” at 66% (1/2 or -200) odds, and bet $100 — so you’d win about $50 if she won the election. Now you come back during her post-convention poll bounce and she’s at 75%; what do you do? You could cash out with a $15 profit, but then what? You could have made $50. Or you could add to your bet at worse odds. Neither is particularly tempting. My guess is you’ll do nothing, and pretty soon you’ll stop visiting the site as frequently.

One way to describe that problem is that the site is trying to get you to think about sports-like betting in a financial market-like framework. But even compared to sports, the idea of waiting months for your bet to settle is frustrating. Most of the betting on sports happens much closer to the event. (According to this chart, 50% of the money bet on an average soccer match is wagered within the preceding two days.)

PredictIt has gotten a lot of things right in terms of basic market structure and fees that previous prediction markets like Intrade got wrong, but I’m not sure how they can solve this basic problem.

One solution would be to mix in political contracts with sports, like they do at Betfair, or with financial markets, as Intrade tried to do at one point. (“Will the Dow close above x?”) But regulators in the US are not about to allow either of those things. Even these political markets require an exemption from the CFTC to operate.

What they’re doing instead at PredictIt is creating lots of arbitrary questions to bet on, so there’s always something new and always something settling soon — like the exact poll average on some random date:


There’s not much trading volume on these questions, because, not to put too fine a point on it, no one cares about them. All you’ve got are the would-be bookies trading with each other.

Let’s go back to our Clinton example for a moment. If you were a perfect rational actor in an economics textbook, you’d go in both times with your exact estimate of her odds of winning, having mentally updated it for events like the convention, and judge the expected value or rate of return of your bet vs. alternative uses of your capital given what the market’s offering, so you’d wind up transacting more frequently before the event itself. You’d also think about the arbitrage opportunities created by closely related markets (will Clinton win, will the Democratic candidate win, will a female candidate win). But very few people really think that way, and for many it’s stressful and unpleasant to be forced to do so.

I’m reminded of another site called Stickk, also started by social scientists, where you can make “commitment contracts” with yourself to quit a bad habit, and attach monetary penalties: for example, you might commit to quit smoking by year end or pay the site $500. Tyler Cowen was skeptical:

I’ve long predicted this won’t work; one group of potential customers doesn’t really want to change, the other group is unwilling to give up control.  It’s not exaggerating to say that human nature is on the line here, and that if I am wrong this is probably the most important idea you will ever encounter.

Prediction markets are running into an analagous problem, and it’s a very tough problem. Even if the novelty of the idea gets people to try it, how do you get them to keep coming back?

Remember those two key factors: interest and frequency. If you wanted to play to interest, you’d let people bet on more non-political subjects. For some of these there are too many people with advance information (Oscar winners, reality show outcomes, who will die on Game of Thrones) and the market could end up spoiling the surprise. For others, like the outcome of high profile trials, there may be considerations of taste or privacy. But from a moral perspective, it’s probably a lot better than focusing on frequency, like lotteries or casinos, and trying to get people hooked on betting on the weather or traffic or some other quasi-random pattern.

I fear that those of us who want to see prediction markets succeed (and I do) have not been focused enough on the core problem: we’re asking people to bet on a subject that 90% of us only care about (in the US) every four years, if even then. Without changing that underlying proposition, I suspect that the current interest in prediction markets, under the klieg lights of a particularly insane election season, is as high as it will ever be.

Graphing Calculators and Competition

This two-year old Washington Post article describes an odd situation that seems to persist today: Texas Instruments is still selling their TI-84 graphing calculator for about $100, though it likely costs only $15-20 to manufacture and they haven’t updated the hardware in over a decade. Phones and tablets can now do all the same things, of course, but they’re not allowed on tests, and I guess you don’t want everyone’s phone out in class either.

At a time when startups are targeting every non-tech business for “disruption,” why has a tech product like this managed to hold on to such a lucrative monopoly? The article describes a cycle in which teachers require the TI models they’re familiar with, parents have to buy them whatever the price, and TI puts a bit of those excess profits back into teacher training and support to make sure they’re still the standard.

No question that’s a big advantage, but it doesn’t sound insurmountable. The competing model mentioned in the article, from Casio, is $80 and running a distant second in sales; they claim they can’t crack that “TI ecosystem,” but my guess is the price difference just isn’t big enough. Shouldn’t someone be able to do this for $40 or $50? Or create a $10 app that locks your phone/tablet in a way that can be monitored or later audited, to eliminate distractions or cheating? (Remember, the TI calculators aren’t perfect in that regard either, since they have persistent memory that can be used for games or cheat sheets, and even the firmware can apparently be overwritten.)

It also reminded me of the RPN calculators made by HP that are still used somewhat in finance, science and engineering. If you’re going to rip someone off, a bond trader is obviously a less sympathetic target than a ninth-grader, but in any case they seem to have maintained the same inflated pricing outside the K-12 education ecosystem.

My first thought was “this is perfect for a big crowdfunding campaign.” A graphing calculator seems like exactly the kind of old-timey hardware device that a small team could design and produce without any proprietary technology, and there’s a social/education angle to the story that should help the campaign spread online. But a quick search on Kickstarter and Indiegogo shows no such projects, and after a little more thought, it’s easy to see why not. Crowdsourced products are infamous for shipping late, and a calculator that doesn’t show up at the beginning of the semester is useless. At the same time, you couldn’t give yourself too much of a time cushion, since your target audience may not even be thinking about this before back-to-school season. And these customers — adolescents and their parents — will be hard to reach anyway, since they straddle the current demographic for Kickstarter and the social networks where the campaigns are promoted.

Why not a venture-backed device startup? It’s probably not a big enough opportunity. According to the article, about 1.6 million calculators are sold per year, and that number has been slowly shrinking as more and more classrooms get school-issued iPads or other new alternatives. It’s still a great business for TI when they’ve got 90% of the market at $100 apiece, but if even if you took half their market share at, say, a $40 price point, your margins would be much lower, you’d need to spend a ton on marketing and support to get there, and when you do, no one’s going to put a tech multiple on a shrinking business.

What about a more developed secondary market? The article doesn’t say what, if anything, TI has done to suppress sales of used models. They’re all over eBay, with almost 10,000 sold in the last couple months (though this may be the seasonal back-to-school peak). That’s still not much compared to 1.6 million total units, though. Here’s one online storefront that sells them refurbished for $83, and another for $84. Again, not the level of discount that would indicate real competition.

Does this represent a gap in the funding market, an opportunity for a bootstrapped lifestyle business or an investor like IndieVC that isn’t built around big exits? What’s the minimum scale for a project like this to pay off? In any case, I’m sure there are a few other examples of obsolete hardware that’s still making 50%+ profit margins, but this seems like a particularly egregious one.

Restaurant Rankings vs. Ratings

This idea seems so obvious to me that I’m sure someone has tried it, but I can’t find any good examples. In a nutshell, it’s an app for user ratings of restaurants that’s based on ranking them in categories, rather than rating them individually on a star or point scale.

So for example, you might have your personal lists of the best burgers, the best Italian food, the best steak, the best vegetarian restaurants, etc. You’d only rank categories where you’ve tried at least two restaurants, of course, and when you try a new one you could add it to your list. So, for example, my list for burgers in NYC might start as follows:

  1. Paul’s
  2. J.G. Melon
  3. Donovan’s
  4. Molly’s
  5. DuMont

Actually, what I’d really have is a ranking of the best burgers I’ve had anywhere, and we could still filter them down to NYC if necessary. But you get the idea.

Now, once we’ve got enough users, we can start aggregating those rankings to present a single ranked list in each category. And this has some major advantages over existing ratings sites:

  • It’s closer to how people already think about restaurants. Everyone has running debates with their friends about their top five burgers, BBQ, or whatever else. Everyone asks “is it better than ___?” No one asks “is it three stars or three and a half?” But sites like Yelp and TripAdvisor ask their users to convert those relative opinions into absolute ones, only to aggregate them back into relative lists (“10 Best Mexican Restaurants in Los Angeles”) — and a lot of useful information is lost in that process.
  • Relative ratings are more useful than absolute ones in this context, because they remove a lot of ambiguity. Is a three-star rating for a Mexican restaurant in New York really the same as a three-star rating for one in California, where the standard for Mexican food is higher? It probably depends on the user. But this way anyone who’s had Mexican food in both places is ranking every restaurant on the same list. (In fact, if you wanted to convert the relative rankings back into star ratings, you’d be more accurate than if users just entered star ratings to begin with — but again, why would you?)
  • It’s faster, more fun and more social than ratings — which means you’d get a different and broader pool of reviewers. Rankings feel like much more of a “curation” experience, more of an expression of your individual identity or tastes. And they still allow for a certain “power user” dynamic without allowing them to be quite as dominant.
  • It doesn’t emphasize rants and raves. Think about how many reviews on Yelp are motivated more by anger or affection for the establishment than any pure desire to inform or help the reader. It means that even with the power users, you only get their input on the best and worst restaurants — these reviews are not a very sensitive instrument for distinguishing one three-star restaurant from another.

This would also be a relatively easy project to get off the ground; you could start with a hundred restaurants in 5-10 categories in a single big city, and a relatively small user base. Even a couple hundred frequent restaurant-goers would provide meaningful data. Then you publish those lists on your site, spend a little money on advertising/promotion and see if anyone cares about them.

I think a lot of people would. There’s a certain appeal to the crowdsourcing and mystery-shopping aspects — which the old Zagat guides really tapped into — but the information would also be more reliable and actionable than anything else available. If a new Szechuan restaurant (my current obsession) opened in New York, I’d honestly be more interested in where it landed in this app’s ranking after a month or two — even if that meant only 5-10 people had visited and ranked it — than in a review in the NY Times or on some food blog.

The algorithm would take a little work, but it also doesn’t have to be as complicated with a small user base in a single city as it would be later — I imagine that cross-city comparisons and aging of reviews would both add some complexity down the road. But that would be a nice problem to have.

Is there a business model? The (alleged) Yelp approach of extorting restaurants to buy overpriced advertising so you’ll remove/underweight their bad reviews doesn’t really work. It would be pretty unpleasant for a restaurant to be ranked last in their category, but unless you’re actively trying to exploit that, you’d probably be showing just the top ten anyway in the aggregate lists — who cares about last place vs. next-to-last?

Could you charge users instead, making this an exclusive/insider thing where you need an existing user’s invitation to join, and pay a small monthly fee to participate? Maybe you could still display your own lists to non-members, but you couldn’t show them the aggregate rankings. (You might waive the fee for any month where someone ranks at least x new restaurants, though you don’t want to give people an incentive to lie and rank a restaurant they haven’t actually visited.)

If you wanted a larger user base, another option is for your ratings to be a loss leader for reservations (looks like Opentable has an affiliate program) or some other commission product, as they are at TripAdvisor and most online travel portals. Even as a completely free service that just plugs into major ad networks, you’d probably at least pay the bills while building up a valuable database.

Now, what else would it work for besides restaurants? It’s hard to think of anything, and maybe that’s why no one has tried this. The startup machine is very good at applying cookie-cutter business models to different categories (so for example, I’m sure there have been a few “Rotten Tomatoes for restaurants” attempts) but ratings are an area where no two categories really work the same way, and most of the other obvious ones have less of a clear advantage for rankings over ratings. There are too many variables on which to rate a hotel, and it would be hard to categorize them. Same with music, books and movies. And with consumer products, most people don’t try enough different ones to rank them; could you tell me your top five printers, cars or vacuum cleaners with the same confidence as your top five pizza places? Better to rely on professional reviews there.

By contrast, there’s one clear variable on which to rate restaurants — quality of the food — and 90% of them fall neatly into a type of cuisine. You don’t need 100% coverage either; if you leave out some fusion or vague “New American” restaurants because they don’t fit into a clear category, so what? There are still a million other review sites that are all over them. And while it’s true that people rate restaurants on other things like service, atmosphere, delivery time, etc, that’s often just as much of a bug as a feature, because sensitivity to these things varies widely and it corrupts the ratings. You could still have people enter text comments on these rankings, and excerpt them like Zagat reviews for the master lists — but they would be more like the brief comments on Instagram posts or FourSquare check-ins, and less like the long, performative stories that seem to rise to the top among Yelp reviews.

The main point is that people care about rankings much more viscerally than ratings. It’s why clickbait sites publish so many arbitrary ranked lists; they know they’ll get lots of comments and shares just from people arguing with them. And I guarantee you some NYC readers have not even made it this far because they were so disgusted by my top five burger choices above. But unlike most exercises in list-making, with restaurants there’s a lot of useful, structured information embedded there. Why not take advantage of it?

The Airbnb Building

By Jeangagnon (Own work) [CC BY-SA 3.0 (], via Wikimedia Commons

There are really two separate businesses on Airbnb and other hosting platforms. Let’s call them “amateur” and “professional”:

  • Amateur: People renting out homes that they live in themselves at least part of the time (either a spare room while they’re home or the whole place while they’re not home)
  • Professional: People renting out homes that they never occupy themselves, for short periods rather than the standard monthly/annual lease

They service much of the same demand, but on the supply side they’ll be governed by different rules and economics in the long run. Let’s look past the current city-by-city political fights and even Airbnb itself (because these businesses will persist whatever happens to one particular company) and think about how the underlying rules and consumer behavior are likely to shake out.

There are two systems of rules in play here, which we might call public and private. The public rules are municipal laws and regulations. The private rules are a matter of contracts – leases, condo/HOA bylaws, insurance policies – and will start to draw more focus once the public rules have been sorted out.

In large US cities, it’s often the case that both types of hosting are technically illegal, but these laws are rarely and inconsistently enforced. Over the next few years, most large cities will settle on a new set of rules with more enforcement behind them, with San Francisco’s recent Prop F debate as one of the first major rounds in this process. I think a reasonable bet is that it shakes out as follows, with one or two cities passing comprehensive legislation and others copying and tweaking it:

  • “Amateur” hosting will be explicitly allowed, with
    • a limit on rental days per year, probably 30-90
    • new transaction taxes (though less than hotel/tourism taxes)
    • new safety requirements (again not as strict as hotels)
    • some type of public registry for hosts, to enforce the above
  • “Professional” hosting will be more or less disallowed, or at least regulated much more harshly

Now, it’s obvious that Airbnb, or any platform that combines both types of host, has an interest in blurring this distinction. Most of their hosts are amateurs, but a professional is worth far more in commissions, because they run higher occupancies and often list multiple units. So Airbnb does everything they can to highlight the more sympathetic amateurs, while pushing for regulation that’s friendly to the less sympathetic professionals.

Ultimately, most voters and public officials will see the difference, and at that point there’s no real public interest or powerful constituency to stand behind the professionals. And the hotel industry, which lobbies against both businesses, is also smart enough to know that the pros are a far greater threat to them than the amateurs and eventually exploit the same wedge.

We won’t argue about the fate of professional hosts, which is really a more interesting question for Airbnb investors than the rest of us. If I’m wrong and that business keeps growing, it’s really nothing new, just an update of the European-style “holiday flats” businesses that have been around since long before the internet.

What’s new, and much more of a social change, is the widespread adoption of “amateur” hosting, and what I want to talk about is how those private, contractual problems – the landlord/neighbor problems – will shake out once the regulatory groundwork has been laid.

These problems are particularly annoying for urban apartment dwellers, and will get a lot worse once those public registries are in place. After all, even if your city allows you up to sixty days of short-term sublets, your landlord can still tear up your lease if they catch you – or if you’re a condo or co-op owner, your neighbors can complain (not unreasonably) that they didn’t buy into a hotel and don’t want strangers around, and get the building rules changed.

So here’s my idea: an “Airbnb building” which explicitly allows short-term hosting and manages it through the building itself.

It could work for a rental or condo building. Either way, when you know you’ll be out of town for a night or more, you’d simply tell the doormen or staff, or enter it into a web calendar, and they’d list it on Airbnb for those nights. If someone books it, they’d arrange for the keys, cleaning and so on. And then you split the extra rent – so for a three night booking at $200/night (after Airbnb’s commission), the building would take in $600 and give you $300.

This may sound familiar — it’s more or less how condo hotels and many other vacation properties are managed. Although in those cases the usage is reversed; the owner is typically there only a few weeks a year, and most of the time the unit’s in the rental pool. In our example, the primary resident is still there most of the time.

This approach neatly solves the landlord/neighbor problem, because the landlord’s getting their cut and the neighbors all know what they signed up for. Most people are still not comfortable renting out their homes to strangers, but you’d expect the people who are to self-select into buildings like this as soon as they start to become available. A typical rental building turns over as much as half of its tenant base every year, so it wouldn’t take long to convert all the leases to allow this; with a condo you might have to sell it that way from construction.

There are obvious time efficiencies and economies of scale in having the building management – who already have photos, floorplans, spare keys, and so on – manage the whole process, as opposed to individual hosts.

And as a traveler, I would hugely prefer to stay in a building like this. I wouldn’t have to worry about whether the neighbors and staff know I’m a paid guest, and the level of trust is far higher. Right now I’d probably never stay somewhere without previous positive ratings, but if I was booking with the building rather than an individual, then just a few positive ratings for any units in the building would make me a lot more willing.

Finally, what if your half of the Airbnb rent was passed on as a reduction or partial rebate of your base rent or condo maintenance fees, rather than additional taxable income? That’s how many amateur hosts think of it already, and the tax savings could be substantial.

It doesn’t solve the insurance problems (liability for theft, damages, injury, etc), and I don’t know enough about that business to speculate, but more professional and centralized management certainly can’t hurt in sorting those out.

Why hasn’t Airbnb set up a building like this themselves, as a kind of prototype to show how well their concept can work? Maybe they don’t want large property management companies to get any ideas. After all, they might get tired of paying those high commissions and just start their own website. In a way, Airbnb benefits from having its hosts operate in an uncomfortable gray area with their landlords and neighbors, because it prevents them from working together openly to gain negotiating power with the platform.

I get the feeling that some newer condo buildings with a high proportion of investment buyers are drifting towards my model already, on the professional side; maybe those buyers had initially planned to rent out their units in standard one-year terms, but they’re realizing they can make much more by listing them on a short-term rental platform and tipping the staff enough to look the other way.

But again, this isn’t meant to be an argument for or against Airbnb, HomeAway or any other particular platform. It’s more about starting from the bottom up: what do hosts and guests want, and what systems and structures can address that in the most efficient way? In the end, the companies that adapt themselves to those structures will succeed. The model I’m describing certainly doesn’t work everywhere, but for serviced buildings in large urban markets (which are already a substantial share of inventory on these sites) it seems like a much better approach. I wonder who’ll be the first to try it.

The Lives of Strangers


This week I found a 100-year-old language textbook in a used book store. Inside were a pair of 40-year-old letters written from a family in East Germany to an American relative:


My Dear Frieda! Many thanks for your lovely letter of July 5th. As you wrote, you still haven’t received a visa [to visit East Germany]. If it really doesn’t come through, then please let us know immediately. As you know, from August 5th to the 20th we won’t be at home, since our vacation to Usedom Island on the Baltic Sea is already booked and paid for. But considering your planned route, you could still visit us before then. It would be great!

Now, you wanted to know, if you don’t get a visa, whether we could come to Munich or [West] Berlin. Unfortunately it wouldn’t work, we’re not allowed over the border. Only pensioners receive permission for that. We can’t go from East to West Berlin either.

We wish you lots of fun on the rest of your trip and hope that we can perhaps see each other again.


Dear Frieda, I’ve inquired again. If you don’t get the visa, there’s another possibility to see us in East Berlin. You can get a day pass, although you’d have to return to West Berlin the same evening. It’s five marks for one day. We could meet in the Hotel Berolina. You’d have to go through the wall on foot, but then you can get a taxi to the hotel. We could meet in the lobby, it’s a nice place to sit and chat. You’d just have to tell us which day you’re coming.

Who was Frieda? I’ve cropped out the last names as a matter of privacy, but with a little googling it was easy to find some of her official records. She immigrated to the US as a young woman in the early 1920s, settling in the same county where I bought the book, and likely still lived there when she died at the ripe old age of 96, after which her books surely sat around in boxes until they were bought by this shop.

She’d have been in her 70s when these letters were written, maybe taking a long-planned retirement trip back to Europe, bringing this old book along to brush up on her German. (The first letter is sent in care of other relatives she was staying with, possibly in West Germany.)

That’s where I stopped digging. I don’t think the letters are so personal that it would be worth tracking down and intruding on her surviving family to return them — just a few innocuous travel plans in a fairly standard American life. But I hope she got to see these relatives in the East.

That same day, my friend Max sent me this post about the disappearance of used book stores. I told him about the letters, and he replied:

Makes me wonder how well we’ll be able to preserve the memories of peoples’ existence as things continue to go digital… [social media] accounts and profiles get wiped at some point, at the latest point upon our deaths. Which means it’s pretty easy to wipe out a sizable portion of a person’s history, if no one is around to preserve the data. There isn’t even a library or rare books collection you can go to to find scraps. I wonder if future historians will be able to rely on caches of email data, or if those will get lost too. What’s the equivalent of all those black and white photos you find at flea markets?

It’s a sobering thought, isn’t it? Today this whole correspondence would take place over email, and forty years from now it could be gone without a trace. There’s no digital equivalent to the way I found these letters, no way to make this kind of brief but moving contact with the life of a dead stranger.

To be fair, it’s not true that all online profiles are wiped as soon as a user dies. Facebook, for example, will freeze and “memorialize” your profile, hiding it from public search results and leaving it open for your friends to view. But that’s only good for as long as Facebook is around. Social networks come and go. When Friendster finally breathed its last in 2011, they gave users less than two months to export their data before deleting it forever.

Modern social networks are better about this, but how many people have really taken advantage of those export tools? And on the rare occasions when users do think ahead about this, just as often they have the opposite goal, to make sure their data will be deleted when they die. Anyway, even if your family does export all your data or keep a memorial profile up, it’s not something that a stranger would ever stumble across.

Digital media are potentially longer lasting than analog media, of course, because they don’t decay (well, barely) with time or reproduction. But the way we treat them is very different. For example, when I graduated from college, I saved all the emails from my school account, and I had every intention of keeping them forever. But over the next few years, as I moved all my files to a succession of new computers and external hard drives, that email archive just fell through the cracks. By contrast, I still have an envelope full of handwritten letters I had received at summer camp ten years earlier. It’s not that the content of my college emails was any less meaningful, just that it was so much easier to lose them through simple neglect.

As always, technology cuts both ways; without the internet it wouldn’t have been so easy to learn more of Frieda’s story. But those birth and death records would have meant nothing to me without the letter, whereas the letter still carries emotional weight even without any outside context.

I don’t mean to get too misty-eyed about this. I realize that of all the humans who have lived on this planet, the vast majority have left no trace at all. But in the 20th Century, the spread of literacy and consumer photography meant that millions more ordinary people left some kind of physical footprint, some testament to their existence beyond just entries in a government ledger. It would be strange if technology is now beginning to reverse that trend.

“Those family snapshots and handwritten postcards at flea markets are heartbreaking,” I replied to Max. “Either the people involved are dead, or they’ve lost them, or they just didn’t care enough to keep them, and I don’t know which is saddest.”

But wouldn’t it be even sadder if they were never preserved at all?