The Bloomberg Terminal Shows How to Create Virtual Economic Cluster

Plus! Big Tech M&A, European Vacations, Reverse Hong Kong, Microsoft-on-Microsoft Disruption, and more...

This is the weekly free edition of The Diff, the newsletter that  tracks inflections in finance and tech. Last week’s subscriber-only  posts:

As a reminder: effective June 5th, the price for new Diff subscriptions will be $20/month or $220/year. Subscribers who sign up before June 5th will pay the current price of $15/month (25% off) or $150/year (32% off).

The Bloomberg Terminal Shows How to Create Virtual Economic Cluster

Serendipitous encounters are the economic engine of cities in a  service economy. You can build a service business anywhere you can get  Internet access, but all else being equal, the best place to build one  is in close physical proximity to people you can hire, sell to, or raise  money from. These economic clusters exist around the world: in New  York, Silicon valley, Tokyo, Shanghai, Hong Kong, Frankfurt, London, and  more.

For some cities, and some industries, this can lead to runaway  increases in salaries and real estate costs: profitable industries pay  higher wages, pushing up housing costs; this prices out more marginal  businesses, but doesn’t strongly price out the dominant one; so as  housing prices rise, the cities network effect improves enough to offset  it, at least for the winners. Silicon Valley used to have a fair amount  of manufacturing, but electronics assembly doesn’t have the same  network effects that hardware design or software do, so that business  got offshored in the 80s and 90s. In New York, finance is slowly pricing  out the media and fashion industries, but those industries can hang on  for a surprisingly long time because they’ve shifted their pay package  away from cash comp and towards prestige instead. Hong Kong also shifted  from a manufacturing-centric economy—they were the world’s largest toy  exporter for decades—to a financial one.

An interesting element of this dematerialization thesis is that the  industry that leaves often leaves behind a higher-value-added remnant.  Silicon Valley became a good place to do software in the 80s because the  hardware it would run on was being built nearby. New York’s finance  industry is, in part, a distant echo of its shipping industry—ships need  insurance, and insurers need somewhere to put their money. Hong Kong  grew as a financial center in part because it was the gateway for  investment into China; a banker who helped a toy manufacturer move from  Hong Kong to Guangdong could help a TV manufacturer move its factory  from Gifu to Guangdong next.

What leaves is a business whose competitive moat is a fixed-cost base  combined with local talent. Factory equipment is hard to move, and  people are hard to move, so moving both at once takes serious effort (or  lots of frequent flyer miles; Chinese companies have started doing more  manufacturing in India, but the factories still have Chinese managers).  But what’s left behind is a more abstract regional competitive  moat: a place where people with certain ambitions know they have to be,  but where they have to realize those ambitions if they’re going to have a hope of paying rent.

There are a handful of entities that have managed to recreate the  economics of service-sector city clusters, without the geographic link:  Bloomberg, BookFace, study groups for challenging courses, private chat  groups, TikTok, and 4chan. They’re all very different: Bloomberg is an  extremely expensive instant-messaging app that also provides financial  data; BookFace is the social network for Y Combinator founders; the last  three are incarnations of social networks.

What they all have in common is that participation is contingent on  expensive, continuous vetting process. For Bloomberg, it’s literally an  expense: $24k/year, with a two-year minimum. BookFace is available to Y  Combinator investees, including alumni. Y Combinator has only removed a few people,  but membership is quite exclusive. Study groups are a harder phenomenon  to parse: for every legendary weed-out course, you can find a) students  who claim it’s not possible to pass without the help of a study group,  and b) students who claim that they didn’t need one. (There’s  something in the water in Cambridge Massachusetts that makes smart  people deathly afraid they won’t be recognized as brilliant.) But study  groups require give-and-take; they filter in anyone who could do well  but not ace the course, and generally filter out anyone bound to drop  it.

And then there are the social networks: TikTok, and especially 4chan,  require fluency in memes. And both products give users a constant feed  of new material, with very little memory for what came before. So to be  an active contributor, you generally need to be a very active consumer.  (Some corners of Twitter have the same dynamic,  but it’s not quite as intense. Private chat groups develop their own  idioms and miniature memes; miss a day or two and you’ll have no idea  what’s going on. (Some of the best chat groups are deeply paranoid about  security, and assume that anyone who’s not contributing—and not  contributing things they’d get in trouble for saying publicly—needs to  get removed.)

All of these phenomena have evolved to make it expensive to get in,  stay in, or both. Which means that at the user level, they have a  zero-trust, zero-permission system for rooting out time-wasters—no  blockchain required! (Although, like many blockchain-based systems, they  need a lot of what looks like wasted energy to keep the system  running.)

City-level economics won’t be destroyed by Covid-19 and its  aftermath, even if there’s a meaningful shift to remote work. But as  long as there’s no vaccine, the network effect will be attenuated. Even  if people keep working in the same cities, they won’t hobnob as  frequently or go to industry events. The network effect will localize:  instead of working for Google in Mountain View to stay plugged into  tech, people will work at Google in Mountain View to stay plugged into Google.

Which means that there’s an opportunity to build city-style economics for a remote-work world.

It’s an odd position to be in: Bloomberg is probably the most broadly  useful and least-wasteful way to replicate a city’s economics—it  requires you to burn cash, not time. So the Bloomberg-of-tech has an  approximate price point, but not a product description. Companies that  generate high revenue per customer from other tech companies might try  to build a network off of that, although high revenue means high  price-discrimination, so you don’t want customers comparing notes.

It’s a start, though. What made Silicon Valley expensive was that it was a dense network of related companies. But part of what makes  that network dense today is that it’s too expensive for other  businesses. Tech companies that benefit from network effects are loathe  to cut them off by imposing a fixed price, but any company that’s  already charging a $x,xxx to $xx,xxxx price per user should ask itself  if those users would be less likely to churn—and more likely to work  together—if they had a digital meeting place.

A Word From Our Sponsors

Here’s an investing secret you won’t hear on Wall Street: one of the best investments right now isn’t  a high-flying tech stock or Manhattan apartment complex—it’s art. Why?  For starters, art has outperformed the S&P by over 180% since 2000  with a correlation factor of just 0.13. Beyond that, the total worldwide  art market is expected to grow by over $900 billion by 2026—that’s a  whole lotta Monets.

But the art market is plagued by a lack of transparency and  historically is reserved only for millionaires. Or, at least, it was:  Masterworks is the only platform that lets you directly invest in  paintings by Warhol, Basquiat, and Picasso—just like buying stocks. They  even have a secondary market where you can trade shares with other  members.

So if you’re looking to diversify away from the markets and want the  security of a real, tangible asset, Masterworks is for you. Act fast and  you can skip the 25,000 waitlist.


Big Tech M&A

Tech companies have announced deals at a faster pace than any time since 2015,  with more—like Uber/Grubhub and Amazon/Zoox—in the pipeline. A few  months ago, the big crisis question was whether higher regulatory  scrutiny would offset attractive prices. Today, it looks like we have  the answer: tech companies have the cash balances necessary for M&A,  while fundraising has been disrupted, leading to opportunities.

Related: now that Indian cell phone network Reliance Jio has raised money from strategic  investors and private equity, Google is considering an investment in a competitor.  The Jio thesis is that they can charge less for data and make money on  e-commerce, which makes them a good strategic fit. The Vodafone Idea  thesis seems to be: Jio’s taken.

Europe Reopens for Vacation Season

The latest on the European vacation macro thesis: Air France-KLM is resuming flights to Italy starting June 1. That’s a little late to match vacation plans—but searches for Airbnb have rebounded to pre-crisis levels. Airbnb supply tends to be short-term countercyclical, at least as long as full-time hosts aren’t blowing up.

On the other hand: easyJet does not plan any Italy flights, due to distancing rules.

Reverse Hong Kong

Hong Kong was ceded to the British Empire in 1842, and spent the next  century and a half as a prosperous, nearly-independent city-state that  benefited from close commercial ties to Britain at one end and China at  the other. As China integrates Hong Kong more closely with the mainland,  some wonks propose a novel plan: found a new Hong Kong in the British Isles, and invite Hong Kongers and Brits to move there.  This is a more plausibly policy than it appears to be at first glance:  the two models of Britain are that it’s run by a) boisterous nationalist  Boris Johnson, or b) technocratic svengali Dominic Cummings. Starting a  charter city and thumbing the country’s nose at China is appealing to  either side. And the first step is already happening, as Britain plans to offer citizenship to visa holders in Hong Kong.

Microsoft Subsidiary Disrupts Microsoft Subsidiary

About half a decade ago, technical hiring managers started to switch  from looking at resumes to looking at Github repos. Today, Microsoft  owns LinkedIn, the biggest collection of resumes on the planet. It also  owns Github. And now Github is testing “personal readmes,”  which sounds a lot like… a resume. This is a nice example of vertical  integration: when the definition of a marketplace changes, companies at  different points on the supply chain scramble to redefine it in their  favor. If LinkedIn were independent, they’d be frantically building a  place to host code right now. But since they both have the same parent,  LinkedIn can gracefully surrender part of the engineer recruiting  market, and Github can cheaply scoop it up.

Airbnb for Warehouse Space

One phenomenon Airbnb benefited from in the 2009-12 period was  duration mismatch: rent is typically an annual contract, but layoffs can  happen instantly. For many Airbnb hosts in New York, and later  throughout Europe, putting a room on Airbnb was the only way to avoid  eviction. Today, renters and homeowners are somewhat less financially  stressed, but retailers are definitely feeling the pain. Enter Warehouse Exchange,  which just raised $2.2m to match ecommerce companies with business  owners who have free space. Warehouse Exchange is pretty interesting as  its own company, but very interesting as a Shopify acquisition.  A distributed logistics network is a very nice complement to a  distributed ecommerce platform.

A16Z on Accounting Versus Economics

One frustration with evaluating tech companies is that the accounting  treatment of costs doesn’t quite line up with economic reality. In a  perfect world, R&D expense would be discretionary—you’ve invented  it, now you can sell it! Marketing expense would monotonically decline  as a share of sales—you’ve sold it, now you’re better at selling it! And  cost of goods sold would be stable. I’ve shared my frustration with  this before:

A16Z offers a more helpful approach, cognizant of the fact that a) margins change over time, but b) they change in predictable ways. The most notable piece to me:

Another  metric we look at is decreasing customer acquisition costs (CAC): is  your cost to acquire each customer falling or staying flat? Combined  with growing organic traffic, flat to down paid CAC translates to  declining blended CAC. While it is common for CAC to go up in  the earlier days of a startup, as you build a strong brand that drives  word-of-mouth, we look for CAC to come back down after you achieve  product-market fit.

That’s a very important point. The two forces at work here are a)  channel saturation pushing customer acquisition costs up (you reach the  point where you’re buying every available ad on your most efficient  marketing channel, and the next-best ad is more expensive), and b)  rising operating efficiency and better network effects pushing customer  acquisition costs down.

How Social Engineering Drives Technology

Samo Burja has a great piece in Palladium,  addressing the deployment problem. Many technologies are academically  interesting, and satisfying to work on, but only economically meaningful  at scale. So the process of growth doesn’t just mean inventing  something new: it means persuading a large number of people to adopt it.  Sales and marketing are lower-status than invention (Watt got a unit in  physics named after him; but I’ve only seen his partner Boulton  memorialized in half of the name of an East Village gastropub).

Samo has written a lot about institutional health and institutional decay. This piece is no exception. The key argument:

Surprisingly,  even organizations dedicated to the creation of new technologies seem  to become hostile to innovation over time. The underlying reason is that  contrarian ideas—as all new technologies are by definition—almost never  survive committees. How could they? By their very nature, they cannot  have the majority on their side. If they do, it is because they have a  powerful champion who has cornered the committee, an uncommon skill.  This simple observation rules out the most frequently proposed reforms  of philanthropy, academia, and government as ways to kickstart  innovation. It opens up new ones, too.

Committees are commonly used in our society because they create the  illusion of avoiding risk. They are a wonderful device for avoiding  responsibility while making the institution seem more rather than less  accountable. Modern institutions have overloaded on actual risk while  fleeing the appearance of it, especially if you count “failing at core  mission” as a risk. Such aversion to the appearance of the unusual can’t  be justified on economic grounds. Rather, it is a socially driven  aversion. There is no immediate reward for making a meeting awkward  either in the boardroom or the engineering room. There’s not even a  reward for making it surprising.

Systematic Value, Re-Revisited

Wayne Himelsein and Mike Green at Logica offer an options-based look at how systematic value investing pays off. Value investing works when value stocks stop being priced as value stocks.  Buying something cheap that stays cheap forever is no way to get rich.  But the switch from value to not-value requires a value investor to  sell, and systematically trading in response to price changes is  theoretically equivalent to an options trade, so you can think of the  returns from systematic value as being returns from an options-writing  strategy. This piece shows that that’s empirically, rather than just  theoretically, true.

How Stripe Scales Remote

About a year ago, Stripe announced the fifth engineering hub: remote. This turned out to be a good call. They have a writeup of what they’ve learned since.  A lot of it should be especially interesting to people struggling  through the first few weeks/months of remote work. Specifically:

One  is the loneliest number. The typical engineering team at Stripe has  five to eight engineers. When only one member of that team is remote,  they often suffer a combination of isolation (both socially and with  respect to work-related decision making) and organizational burden  (because they are effectively responsible for rearchitecting the  team’s processes to be remote-friendly in addition to doing their  actual job). Instead, moving multiple remote engineers  simultaneously onto a team has yielded much better results for their  productivity and happiness. This shift acts as a forcing function to  support asynchronous communication, better distribute the workload, and  accelerate the adoption of team norms to socially include all members.