it is remarkable that internet markets with low barriers of entry often settle on monopoly/ologopoly models so quickly and easily.
Note: I am going to pick on this statement, which is one small part of a comment that otherwise makes good points.
I think that there is a significant barrier to entry on the Internet that is rather subtle. Incumbent firms know a lot more than potential new entrants, and the Internet enables them to exploit this advantage quickly and at low risk.
Suppose I am managing the incumbent firm X, and you want to enter the market with firm Y. In textbook economics, Y just has to copy what X does. But that does not work in the Internet context. Most people say that is because of network effects and lock-in, but I think there is a more important reason.
The problem with your copying me is that by the time you finish copying the current version, I will be two or three versions ahead of you. The thing is, at any point in time, I have a better idea than you do about what is sub-optimal about the current version of X. So I know which features I am going to drop, which features I am going to tweak, and which features I am going to add in my next version. You may have some guesses about all that, but you don’t have the experience that I have.
The knowledge that managers accumulate while operating a business becomes a big entry barrier. When I was the “chief scientist” of an Internet-based business in the 1990s, we were periodically spooked by VC-backed competitors who had enough capital to bury us–if they had our experience. But time after time they squandered their money trying stuff that we already knew didn’t work. One of my partners described it as like being the villain in a cartoon car race. We are the lead car and out the window we’re tossing obstacles for the other cars: nails, oil, what have you. We were engaged in “move fast and break things” before that slogan was invented.
In traditional businesses, followers tend to be more agile than leaders. In today’s Internet environment, it can be the other way around. And there is nothing that fends off competition as rudely as a market leader that is able to move faster than a follower.
As I pointed out previously, this just doesn’t explain the dominance of a communication platform like Twitter, which changes slowly, is easy to copy – indeed, has been copied, several times. And when it does make changes, those are quick and easy to copy too. And yet, clearly there is strong lock-in due to network effects.
How should we distinguish between Twitter-like dominance and Villain-car-like dominance?
Twitter isn’t worth that much though, its a mistake to think of them being a monopoly. They are a subset of the communications industry, and they aren’t dominant in that sector, just a complimentary company who does one thing.
That’s like saying a national beer monopoly isn’t a monopoly because after all it’s just part of the beverage industry, complementing all the other beverages. Which would be especially perverse if the rest of the beverage industry consisted of a soda monopoly, a milk monopoly, a water monopoly, etc.
Yes, there’ s a conceptual problem when it comes to differentiated products (i.e., not fungible commodities). The Coca-Cola company does not have a soda monopoly, but it has a ‘monopoly’ on selling branded bottles of coca-cola.
“That’s like saying a national beer monopoly isn’t a monopoly because after all it’s just part of the beverage industry, complementing all the other beverages.”
No it is like arguing that someone has a beer monopoly and basing the argument on the facts that any ‘competitors’ don’t hew to a specific interpretation of the Bavarian purity laws and so technically aren’t beer but barelywines or are other malted beverages.
You only get to “twitter is a monopoly” if you start drawing pointless lines and arguing that instagram, pinterest, facebook, yahoo groups etc are all monopolies as well.
“The Coca-Cola company does not have a soda monopoly, but it has a ‘monopoly’ on selling branded bottles of coca-cola.”
Which is why it is a good idea not to select a definition of monopoly that has no real meaning, because technically every company with a brand name then becomes a monopoly, giving us nothing of value in the word.
There’s another way to go about answering the question. When we look at some foreign version of the internet platforms which are dominant in America, and which have dominance in their own national or linguistic infospheres, so we see them competing head-to-head with the US versions, with the more agile, villain-car firms always winning and staying ahead, or do they seem to enjoy the same kind of lock-in and network effects in their regions? Yes, there is the trouble of some of the companies being state-sponsored / de-facto monopolies, and of those states putting US firms at a disadvantage. But actually, several of the Chinese versions (WeChat, Weibo) seem to have superior ratings, but don’t cross over. Japanese Line is definitely just as good as WhatsApp (I’ve used both), but each app doesn’t tend to penetrate the home markets of the other.
Furthermore, let’s put some equally ‘agile’ and sophisticated companies head to head. Aren’t Apple, Facebook, Google, etc. all really, really successful at quick, quality software development management? Yet when Google has tried to take Twitter and Facebook on, merely having quality products was not enough to overcome the other companies’ first mover advantages, and those projects were failures.
Meanwhile, where don’t we see monopolies? Webmail is a great example. Gmail is a great product, but there are still huge numbers of people who use Microsoft, Yahoo, Mail.com, and a bunch of foreign providers. This despite the fact that many of these email services are clearly inferior to Gmail. This fact illustrates a form of inertial ‘lock-in’, but also a lack of concentration and centralization. But it’s not even really ‘lock-in’ since you can use other applications to interface with one’s account, and really, it’s just about the preference for keeping one’s email address which is the same as wanting to keep one’s phone number or maiden name for professional recognition reasons.
Why don’t we see centralization in this case? No network effects. It doesn’t matter what email service you have, because your accounts talks to everybody else with any other kind of email address, without discrimination. Same goes for phone communication. At a first approximation, anyone with any kind of web browser can real anybody’s blog no matter what blogging software they use, so there are lots of competitors at various levels of ‘agility’, quality, and software development sophistication.
As soon as one loses easy interoperability over common languages and protocols, boom, centralization into a de-facto monopoly across the home infosphere.
“it is remarkable that internet markets with low barriers of entry often settle on monopoly/ologopoly models so quickly and easily.”
Can anyone come up with an example where the market doesn’t have obviously steep network effects?
Google search. There’s no intrinsic network effects in search. It’s dead easy to switch search engines. Yet Google has consistently been outpacing its competitors. Arnold’s explanation is as good as any for this.
I think that’s a legit example. You/Arnold may be right.
I wold argue that this is a market that has very, very high barriers to entry. Yes, the barriers on the demand side to switch is very low. But on the supply side, deployment would require billions in infrastructure at this point to compete. Providing fast, complete searches requires indexing the entire internet, and returning answers quickly.
I’m a bit surprised that Amazon hasn’t taken a shot given that they are 90% there already.
An alternative strategy would be to outcompete Google with more tailored solutions in specific subcategories. The greatest danger to Google may be death by a thousand cuts.
At this point, a team with a better algorithm would probably shop it to one of the giants instead of going up against Google directly.
A good subcategory would be, eg, computer/software tech. The issue is, subcategories like these are already owned by websites. Every techie knows that Stack Overflow in a search result is where you would start, so why use “techsearch” when google (or bing) pops up the result you want?
Yes and no.
The knowledge that managers accumulate can be a barrier to entry but I’m not sure the post has proven that the Internet is comprehensively different in this respect than many other businesses. For example, a large part of Intel’s dominance is due to its manufacturing abilities which is an example of managerial expertise and is distinctly not-Internet-flavored.
I would argue that the Internet actually makes managerial experience often less of a differentiating factor because it can force-multiply acquired knowledge. For example, a single SEO genius or distributed systems architect can apply his or her knowledge across an entire organization and give a company huge gains from a single hire in a relatively short time. Compare that to, say, an expert in manufacturing lines whose ability to effect change will be generally limited to one plant at a time. Hence, smaller firms who can hire right can staff up and compete with the larger guys relatively quickly.
That’s one of the real differences today vs the 90s. Back then the Internet space was so nascent a lot of people didn’t really know what they were doing. So companies ended up having to reinvent the wheel over and over, like building payment systems. Today, most of the “base-level” knowledge exists everywhere – instead of building a payment system, you integrate Stripe. So the surface area for being a car-villain is limited more to your product and not all the things (marketing, payments, security, infrastructure, etc) that used to differentiate companies.
I’d guess, that because of government regulation, that telecom and pharm industries have the leaders as more agile than the followers.
I’d argue that with internet companies the ability to scale up fast is unprecedented… so a still quite small group can have worldwide impact, while retaining the ‘move fast’ capability. Also seems to exacerbate the ‘winner take all’ outcome. But occasionally a leader leaves an opening that a new entrant can exploit (think MySpace). And we’ll see how these companies behave as they get older and larger – as noted, Twitter does not seem to change things very fast.
This doesn’t explain why senior (or middle) management don’t leave to start their own venture. The rewards are significantly higher.
Effem’s point seems strong to me. Technically it should be easier to start rivals in the new programming-silicon Silicon Valley than it was in the old etching-silicon Silicon Valley.
There are admittedly truly ridiculous economies of scale in software, and many of the products have large network effects. But the pattern we see doesn’t seem to quite fit that.
As I understand it, Twitter and Youtube don’t have a very good revenue model, so it is tempting to think of them not as profit-making businesses but instead as PR/lobbying/campaigning or vanity projects akin to various influential newspapers. That might be displaced by someone else with deep pockets and comparable motive and freedom to operate, but it’s not likely to be displaced by an ordinary hungry business competitor.
Also, there has some decay of rule of law along the old startup growth pathway, in ways that seem likely to promote the kind of cronyist sclerosis we associate with aggressively mixed national economies, and with heavily mixed segments of our economy such as last mile telecom, pharmaceuticals, and universities. For example, going public under Sarbanes-Oxley apparently has not only a lot of obvious overhead but various nonobvious regulatory judgment calls. AFAIK, every single technically successful large SV firm is horribly vulnerable to discretionary enforcement of HR standards such as disparate impact, and so operates only by informal exceptions. And those firms that do have revenue models tend to be critically dependent on banking services — esp. credit cards — that have been openly targeted by formal government leverage such as Operation Chokepoint, and also sure look as though they are informally targeted at the behest of powerful political and cronyist interests in more recent rounds of shutting out the NRA, and in the “slow walk it to regulatory death” recently described in https://johnhcochrane.blogspot.com/2018/09/fed-nixes-narrow-bank.html . It would be unsurprising to me if this sort of change in the governing institutions led to the usual pattern of sclerosis in the mixed-government “business” institutions that correspond to the older business institutions.
Also, it is suggestive that SV firms have reportedly gotten a lot more aggressive about spending money on lobbying than they were a few decades ago, especially before the big Microsoft antitrust battles. I think probably reasonable people can differ about how large the effect is, but do I have a strong opinion about what the sign of the effect is.
The internet may lower entry barriers, but it increases economies of scale in many things, like (as Arnold points out) institutional knowledge.
Try thinking about it this way: suppose shipping costs were zero. Wouldn’t this tend to concentrate the industry of making stuff, because there are no longer geographic barriers to consolidation? When the whole world is one market, instead of many local markets, how can that *not* lead to more concentration?
That only holds if the movement costs of unlimited amounts of all the other inputs it also zero (to include specialized labor, or special qualities of the geography of the production site), and that there are no important time lags. Basically, “global teleportation of everything is free”. When we’re talking about bits, or small, durable, valuable items with very low marginal costs (maybe ‘bit-like aroms’) that’s close to being true already. The further away one moves from that situation, the more opportunity for local bottlenecks, bidding up input prices to the post where the output price is equal to that offered by at least one other firm.
Even for delivering bits, it’s worth pointing out that firm concentration can be much more concentrated than the infrastructure moving and processing the bits themselves, which is highly distributed geographically due to flow rate limitations, (e.g., For tons of data, you can still sometimes FedEx hard drives quicker and cheaper than transmitting the information digitally.)
The problem with your copying me is that by the time you finish copying the current version, I will be two or three versions ahead of you. The thing is, at any point in time, I have a better idea than you do about what is sub-optimal about the current version of X.
Except that the tech world is full of examples where firms dominant in a particular market segments were overthrown by late-comers. Many people think Apple created the smartphone industry when it introduced the iPhone. They’ve forgotten that there was a robust, rapidly growing smartphone industry for many years before that — one dominated by Nokia, Palm, RIM/Blackberry, Microsoft, and Palm. It’s funny to read this now, with none of these companies having any idea of the creative destruction coming their way:
https://www.theregister.co.uk/2004/10/27/pda_market_q3_04/
Apple was a late entrant. But even later was the ultimate dominant player Google with Android. Google, in fact, was a late entrant not only with Android, but also with its original search product (following Altavista, Excite, Yahoo, and others) and mapping (remember MapQuest?). Google Docs came decades after Microsoft Office, and Chromebooks came long after Windows and Mac PCs and even iPads, but that isn’t preventing Chromebooks from rapidly taking over the education market. And, of course, Facebook was a late entrants to the social networking space (coming after MySpace and Friendster).
One reason this happens is that there are often distinct disadvantages to being several versions ‘ahead’ — you are technologically and organizationally locked into a particular approach. So consider the Chromebook example. What’s great for schools is these computers are very inexpensive and amazingly cheap and easy to manage. No software is installed. A student can pick up any school Chromebook, sign in, and be ready to go with all their documents and applications. Yes, Microsoft and Apple had a long head-start in education with years of experience and established relationships and sales networks. But being ‘3 version ahead’ left them unable to match the cost and simplicity. They would have to re-architect *everything*, and if they were successful, what they’d win is the right to sell products at a fraction of what they were used to. They clearly didn’t have a better idea of what was sub-optimal in the existing systems (just as all of the leading 2004 smartphone vendors did not).
It kind of feels like there’s an unstated assumption in this post: It should be easy for newer, smaller companies to disrupt big businesses. But the Internet is special, and gives advantages to big Internet companies.
I’m not sure this is true in general. Yes, big businesses become ossified, and slow to react, and are then often caught flat-footed by more nimble companies. But in their prime, big businesses have always been monsters to compete with. Would you like to have gone up against McDonalds or Walmart or General Electric in the decade or two when they were dominant?
I think the big Internet companies are still in their “prime” stage, and haven’t yet aged enough to reach the “ossified” stage. Now, if the idea is that they’ll never become ossified as they age, just because they’re Internet companies, I feel doubtful.