Winner Takes Most

The history of the Internet and mobile is that in many categories the winner takes most of the market:

  • Search – Google
  • ecommerce – Amazon
  • Social – Facebook
  • Ridesharing – Uber

We can go on and on with making a list like that and I have left off many many markets, but I think this short list I made at least gets the point across.

The reasons are many, but at the core are network effects and the fact that the more users and data a service has, the more value it can create for its customers and users.

We strongly believe in network effects at USV and look for them as the primary form of defensibility in the investments we make. We don’t always get things right and we certainly don’t always end up investing in the company that wins the market. But we understand how these things work and invest with the mindset that winning a market can result in a very large return on investment.

Lately, we’ve been wondering if there is an end to this pattern on the Internet and mobile. We think it is possible that an open data platform, in which users ultimately control their data and the networks they choose to participate in, could be the thing that undoes this pattern of winner takes most. The blockchain is the closest thing to emerge that looks something like that. But the blockchain hasn’t (yet?) shown that it can produce something important like Google’s search or Facebook’s social graph and until it does, we are just waiting.

This is an issue for society to ponder. As I have spent time in Europe this past month, it’s easy to see that the search engine they use here is Google, the social graph they use is Facebook, and so on and so forth. If the US produces the networks that win most of the market, that’s an issue for the rest of the world. The Chinese have dealt with that issue by protecting their market. The rest of the world (mostly) has not.

Will that always be the case? Will the countries with the most sophisticated tech startup communities end up winning the global economic race as we transition from an analog to a digital world in which the winners take most of the market?

It’s unclear to me how all of this plays out, but it’s been on our minds at USV and we are talking a lot about it. So I figured I’d talk a bit about it here too.


Comments (Archived):

  1. kjetil

    I believe you’re wrong when saying that Europe = “the World outside the US”. EU/US should in my opinion almost be seen as one market (EU countries are almost like US states with more regulations), but the rest of the world is not adapting the same technology.Take the messaging market for instance – where FB messenger takes the EU/US market, but others take the market elsewhere.

    1. jason wright

      yes, but European directives from Brussels have the power to override local regulation and law. that’s the democratic ‘deficit’ that right wingers in the UK are so upset about, their loss of local autocratic autonomy πŸ™‚

        1. JLM

          .Good humor, well played.JLMwww.themusingsofthebigredca…

        2. sigmaalgebra

          Gee, two days ago I put some olive oil in an 8 quart pot, added onions I’d cut my way, canned, peeled, whole tomatoes, cut up and drained, canned crushed tomatoes, garlic, oregano, basil, parsley, and bay leaf, simmered. Gee, in the EU they might send me to jail for doing that!

        3. Vasudev Ram

          That was a good TV series. I had seen some episodes.Who says the British don’t have a sense of humour?

    2. fredwilson

      You are 100% correct and I was not saying that Europe is the rest of the world. I was just saying that being in Europe made me think about this issue.

  2. jason wright

    a (“the”) network effect exists within a network created by a particular set of technologies ordered in a particular way. as the technologies change to create a new order of network a new effect is observable.”the” network effect of today will eventually be known by another name to distinguish it from a new effect in a new network. Network Effect 1.0?i like to think of money and bank accounts. it’s my money and i can choose to put it in any bank i choose, and in any currency i choose, and share it with whomsoever i choose. my personal information and identity should be a little like that, but not held in an account by a third party. it should be held securely only on my smartphone, just like some bitcoin wallets work. as personal devices become more powerful the possibilities for a new network come ever nearer. imagine a network made up only of super powerful smartphones, ubiquitous wireless connectivity, and a software ‘data flux’ membrane.

  3. LIAD

    Not sure understand subtlety of the questionIs the issue that if decentralised applications take off the network ownership piece in terms of value dissapates? And if so, what can be done to earn outsized returns?No value moving down the stack, other options, horizontal added services or moving up the stack. Is that where we’re at?

    1. jason wright

      the architecture of the vertical stack collapses?

      1. LIAD

        No but moving down stack to infrastructure is a road to nowhere in terms of bang for investment buck.I think.

    2. fredwilson

      I do not think we are anywhere other than the network era right now. I am pondering out loud about what may come next. And if its an open data layer then yes value from owning a network will dissipate

      1. boteman

        “The network IS the computer.”

      2. LIAD

        Roger Captain.Framing VC as an exercise in predicting the future actually makes it sound quite compelling.

      3. obarthelemy

        I think legislation is more likely to disrupt than open data: legislation will be enforced, open data requires incumbents’ participation or displacement.Legislation is likely to happen in the mid-term because of monopoly concerns, and of security issues. And of $$ rewards to politicians for getting involved ^^

  4. Aviah Laor

    Financially, the end of this pattern means, somehow, a new model that will replace the economy of scale and a different structure of the “shareholders value”. Legally, it means companies that don’t want to control the servers, the tools, the API, the platform, the data and the terms of service. Technically, is means protocols over platforms. Culturally, it means most people will prefer the unique, where everybody else is **not** going. Finally, figure out how to somehow make money in all these pre-conditions. If this works, it’s probably a good thing.

  5. awaldstein

    Isn’t there another step where some of these platforms reach their limits some don’t?Facebook is certainly social but it has failed at leveraging that to a commerce platform as hard as it tried. They’ve refocused and they are deepening because of it.Uber is still pushing out replacing ride sharing with transportation. They are going to own the package piece and mop up all the little startups like Door Dash who are all going after the 20% of trips which are simply to move stuff around.You are way ahead of me as especially in the personal logistics platforms we are still very early.

  6. David Semeria

    We also spent a lot of time thinking about this, and came to the conclusion that bottom-up analysis is more useful than top-down. With top-down analysis. you try to work out what is special about certain markets that causes them to become so concentrated. With bottom-up analysis you think about how users behave.So taking the example of search, users prefer to download one app, or remember one website, which can satisfy all their search needs rather than using niche sites and apps for different use cases. The same goes for ecommerce, social networking, etc. Nothing too mind blowing there.It’s the next step that is interesting: you can apply the same “path of least resistance” logic to apps and websites themselves. Based on our (informal) research, people are still downloading apps and trying out new websites, but they aren’t making it into the “core set” anymore. In other words, over the past couple of years the web sites and apps people use regularly has become pretty stable.If confirmed, this finding will turn the consumer internet upside down.

    1. pointsnfigures

      What if artificial intelligence starts to do things algorithmically for you? Suppose the Algo is written in such a way that it doesn’t automatically call Uber for a ride, but chooses between several ridesharing apps and figures out which one would be cheapest with the shortest wait for a car? Suppose it doesn’t use the same search engine every time etc?

      1. David Semeria

        Perhaps. But what I’m saying is that people will try the new AI app, say “wow that’s cool” and then carry on using Uber because it’s already on their home screen.

        1. awaldstein

          This thinking, which I agree, is already playing the local delivery world there a zillion options. Every possible vertical silo of topic.Everyone is a pain as it requires new accounts, more costs to the partners and funky inventory control.Uber Rush–well, everyone both has an account and it respects the importance of the partners brand.Simple but non trivial.

          1. David Semeria

            That’s exactly the reason I said the web could be turned “upside down”, Arnold. I think new functionality will be increasingly pushed into the market from the top, rather than from the bottom, as was usual.In other words, the established brands will leverage their financial resources and, above all, their presence in the “core set” to enter new markets. Uber Rush is a great example.

          2. awaldstein

            This is playing out for certain.And really changes how you think about innovation, where to invest your time and innovation.Brand has never been more challenging to build yet more powerful once real.

        2. Richard

          In contrast to Freds prediction that apple will not be one of the top 3 tech companies by 2020, if AI does (and it looks like it will) play out the way you say it will, it will be Apple who us in the best position to take on additional categories.

          1. David Semeria

            Not necessarily. Apple is an ecosystem, not an app. I think Google, FB and Amazon are best placed here.

          2. Richard

            Is siri an app or the ecosystem ?

          3. David Semeria

            Well, Siri is certainly an app, and a potentially very powerful one. It’s just that I don’t see very many people talking *to* their phones. Perhaps that will change….

          4. Richard

            AI will minimze this.

          5. David Semeria

            I you believe that, then bet on IBM. Watson is several years ahead of the competition, and is already available for commercial integration via Bluemix.

          6. Richard

            Does anyone under 30 even know what ibm does?

          7. Dave W Baldwin

            With @hymanroth:disqus on this if you add Google. SIRI is the first round and is narrow. That strategy is about to be placed in the toy market next month (and will fail). A person does not buy the iPhone to get SIRI. Google Voice is better.Basically, you have to think of the next round of AI which will be algorithms that start to become “flexible” which will move AI into a more autonomous state. That is what @pointsnfigures:disqus would be referring to. I was asked by one who reads this if I was going after something on this level (the personal assistant serving the customer, not be under control of phone company) and I couldn’t figure out how to put this out there where Apple, Google, Micro or whomever would allow it on the phone.For now, IBM is a player but remember Google will beat Apple in this game. Another factor is Ng looking long IMO.

          8. Drew Meyers

            I’m 33, and i really don’t know what ibm does. I’d still say build computers. But I just sent to their home page, and they message cloud. So they are the cloud for big corporations?

          9. obarthelemy

            They do services and big iron for large corps. That’s not growing so they’re trying to move into cloud.

          10. obarthelemy

            I think talking is just a first step. Google Now also offers proactive suggestions on my home screen via a widget (a true widget, not a notification Γ  la Apple), on my watch… There’s a wide choice of imput and output mediums, and for AI algorythms and data collection…I’m sure notifications and audio alerts are coming (the filters will be fun: “notify me when more than 100 people die in one stroke in France, a kid gets kidnapped in Marseilles, or the Galaxy Tab Pro 12.2 drops below 300€” ^^)

          11. David Semeria

            There’s a saying in AI which goes along the lines of “More data beats better algorithms”. This is why Google and FB have a big advantage here. In the recent film “Ex Machina”, the android’s AI was taught by search engine queries…

          12. Dan

            Is Siri still powered with tech from Nuance or has Apple moved to their own tech?

          13. Dave W Baldwin

            Not necessarily FB because they copy. It is hilarious to read of their “AI Lab” and it is just a rewording of Deep Learning.

          14. David Semeria

            I believe FB’s willingness to copy features and ideas from others is one of the secrets of their success.

          15. Dave W Baldwin

            So true πŸ˜‰

        3. obarthelemy

          Can’t we make a parallel with Google for web sites ? Most people around me don’t even type URLs, they google the website name (or, quite often, they… google the URL…)If done right (a big IF), I think some broker could insert themselves between services and users. I’m sure services won’t let that kind of control go as easily as websites did though.

          1. sigmaalgebra

            So, a Web site has an IP address. Well, typing those, even just as dotted decimal, was too tough, so give the site a domain name and the Web pages URLs and let DNS servers look up the IP address. Gee, that was too tough, so in HTML display just a short description of the site with the full URL used but not shown. Now we wan to use Google to find the URL so that a DNS server can find the IP address?Gee ….Another approach: I have a dirt simple HTML file with my favorite URLs, with little descriptions of each. My Web browsers always start that file. Since Firefox highlights the sites visited recently, that’s file with its highlights and usual Firefox function to find keywords in a Web page work for nearly everything.For more I have a file of little things I want to remember with key words for each entry. I use that with my favorite text editor with some little macros I wrote. So, the URL at YouTube of a Bob Hope movie is in that file. So, using one or a few key words, I can find the entry, and with the URL on the current line, one keystroke runs Firefox on the movie.For more, if I’m reading a news article and regard it as valuable then I save a copy, index and abstract it, and save the URL. Then with my editor again, one keystroke will open the file, either from the original on the Internet or the copy on my computer.If an article has some URLs of some potentially interesting content, then I do the usual left click and see the article. If it is any good, then I save it and then use it as just above.For each specialized topic, I have a file of URLs, tree names, etc. and, again, one keystroke lets me read the relevant content.So, so far I’ve never typed a URL! I nearly never type a URL or go to Google to get a URL!

          2. obarthelemy

            Sure. Please do come and teach my 80 yo dad, or even my 50ish teacher sister, or even my 25 yo artist niece how to write scripts and juggle URL databases. Then re-explain it to them ad vitam…Personally, I gave up at “Google’s search box is not the adress bar”. After a few tries ^^You’re obviously brilliant. Now go make others brilliant too… good luck ?

          3. sigmaalgebra

            > You’re obviously brilliant.Gee, I’ll try to use that to pay for my next tank of gasoline! Somehow I doubt it will work.Computing has a really big, ubiquitous problem — the whole field, from companies down to computer science professors down to developers, etc., just don’t make clear what the heck they are talking about. The whole field is infected with a terrible disease. In other fields we would call that illiteracy. E,g,, even to use “address bar” is more explicit and articulate than 99 44/100% of the field. To make the situation worse, the JavaScript programmers at The Washington Post have converted the address bar to something else so that have to click around a few clicks to get the URL of the page. How to learn that? The usual way in computing — experimentation. That sounds like an absurd joke, but the approach was explicit in some of the culture of Xerox PARC and their conception of the graphical user interface.If people need help in finding URLs and using them without typing them and want to use Google, Bing, etc. for that, fine. Don’t bother with a new, separate, less general tool when there is already a tool that works.But, for finding content on the Internet, there are bigger needs than helping people avoid typing URLs. Gee, with “on the Internet” I was a little redundant since we are well on the way to that being about the only source of content.

        4. LE

          Also important is the degree to which Uber “satisfices” the needs people using it’s service. They don’t even have to bat 100, just in the 90’s.For example I always use Amazon because it’s good enough to get me most things that I need and it has become a habit. And it’s predictable with returns and I can have all of my purchases in one place and so on.

        5. obarthelemy

          But then you get Reward Miles and discounts and low-price guarantee and … for using the meta-App, which itself is also making money on the other end.

      2. awaldstein

        I’m with David on this one.Behaviorally, especially where the action involves logistics and real life use cases like Uber, people go to a comfort zone. They go to what already exists within their scope.Brand matters as it is the truest label of trust and that tied to behavior is what drives not only adoption but use.Uber is swallowing the transportation market for just this reason.

        1. Richard

          Yep. So long as the categort winner makes the best product there is little room for the artisian baker

          1. awaldstein

            Been an eye opener watching Luli bring in these services to their storefront.Tons of catalog based vertical delivery services, most charging the vendor around 20%.Then there’s Uber, flat rate no brainer.

          2. Richard

            Amazon prime now will likely prevail in this category. Its in LA and is mindblowing good

          3. awaldstein

            HmmmPrime as in delivery on demand from warehouses. Or Prime as a pure logistics infrastructure moving hard goods from place to place.I thought it was the former. Certainly is in NY.Uber is great cause it does only one thing, moves stuff around that it doesn’t own, doesn’t take a charge for.

          4. Richard

            Prime now does both

      3. LE

        Algo is written in such a way that it doesn’t automatically call Uber for a ride, but chooses between several ridesharing apps and figures out which one would be cheapest with the shortest wait for a car?(With search engines there was metacrawler [1] which tried to do a variation of this).I want to agree with this but in the end my brain wants to use what is comfortable look and feel wise. Using your “doesn’t use the same search engine every time” as an example if I do a search and I am used to the way that google formats the search because that is the search engine that I am used to (visually) I don’t want to see results as formatted by other search engines. That’s part of the artifact of brand loyalty people also want to see visual cues which to their brain mean “predictability” and is a comfort zone.[1]

        1. JeffreyKaine

          I think the point here is that you wouldn’t use the interface of any of these apps. You would tell the AI where you needed to go and the AI would do all of the logistics in the background, choosing and booking the service that will get you from A to B the fastest/cheapest/best way possible.

      4. Matt Zagaja

        Siri already does this for pulling content from various streaming services. Not too far off.

      5. Aaron Fyke

        I was going to ask if you had ever heard of “Dogpile” which was a search engine of search engines. Then I was shocked to discover that they were still around! I think branding is a big thing here. If Uber or Google are “good enough”, people won’t switch, even if an algorithm would be better.

        1. Realist50

          Ok, but what if a competing on-demand ride service comes in to some markets (or a market) with the idea of undercutting Uber on price? It seems like an obvious enough to way to try to compete. Uber takes 20% of the fare. New competitor instead takes 10%, offering somewhat lower fares to passengers and somewhat higher fees to drivers. And I don’t need to try to do this in every market, at least not at once. I just start out competing in one or two large markets.That’s rather different than taking on Google, which users don’t pay to use and where the competition would have to be based on adding features or improving results (both areas where Google has large incumbent advantages due to its massive scale).

          1. Douglas Crets

            But existing ride share services already do that. Didi in China, for example, is well below the cost of an Uber ride, and they are already competing in Uber’s biggest China market, Chengdu. The Uber marketing story seems to be about growth, but their competitors, I can assure you, are winning on cost and entrenchment, trust and sponsored-support. Uber is not going to take over China. At least, not in the next ten years, it won’t.

          2. Realist50

            We may be making the same point. I was responding to the claim that customers won’t switch if Uber is “good enough”.My point is that a lot of customers likely will switch if some other service is more or less equal but a little bit – say 5% – cheaper. And the very success of Uber’s service will lead to a greater propensity to switch for relatively small savings. If I use Uber once a month, do I really care if I could save a dollar or two? Probably not. If I use it a few times a week, though, that 5% lower price from a competitor adds up.The logical development here is something akin to Kayak that can check multiple ride sharing apps on a consumer’s phone and present the cost and time of each option. I think it’s an open question whether that would be a separate app, or just a feature that’s built into the phone’s OS.Similarly, this competing service can recruit drivers by offering a slightly higher take (net of commissions) than Uber does. It’s not a big commitment for the driver – just add another app, but then use Uber instead if that other service doesn’t work out. I already encounter drivers who use both Uber and Lyft apps.The above is why I’m extremely doubtful that Uber will end up justifying a $50+ billion valuation. I don’t see how its business has enough barriers to entry to maintain pricing power. I think that margins will be competed down to a low level in most major cities. Oddly enough, what Uber may need – after fighting taxi regulations in so many cities – is regulatory barriers to entry that keep margins up by making it difficult/expensive for new providers to enter its market.

          3. Douglas Crets

            I think there are two threads here. There is whether Uber or any other car service can survive long term against each other. Doubtful that Uber can do that solely by being a car ride provider. Which brings me to the second thread. Uber is valued so highly because it’s not always going to be in the business — exclusively — of delivering riders. It’s a payments app, and it’s long term potential is as a payments provider, and a consumer shopping intelligence tool. Its algorithms know so much about where people go, what they do, and for how long and when they don’t do it. It’s basically taken mapping and delivered it to consumer payments, which gives a Google-like layer of identity and consumption analytics.

          4. Realist50

            Interesting thought, but I don’t assign much current value to the latter.It’s payment system is based on credit cards. Why do third-party merchants and consumers want to get Uber in the middle of a card transaction instead of just dealing with each other directly? That’s particularly the case if the idea is that Uber is somehow getting a take on the transaction.As for its analytics data, I wouldn’t oversell what it has. Uber knows where people go for the trips that they take with Uber, which represent just a small fraction of trips for almost all passengers. My best guess is that it ends up disproportionately being a lot of information about people going to and from restaurants and bars.Maybe Uber can someday make something of these, but I go back to my original point – I don’t see why an investor would assign much current value to them today. As a customer who likes the service, I also hope that Uber doesn’t mess up its core service by pursuing adjacent revenue opportunities to justify its high valuation.

      6. da Tyga

        What you are suggesting is like WebJet, Wotif, etc.Apps that perform aggregation could leapfrog existing gateways. Siri v5+ might be eerily able to do something like that.

      7. Realist50

        I’m not sure it even needs to be that complicated, though that might be where things head. Look at the options that are out there for comparing airfares. Why can’t a service be like Kayak for local on-demand ride services? That could make Uber and its competitors look a lot like the airline industry – a widely-used transportation service that generates a lot of revenue but without a lot of sustained profitability.

        1. pointsnfigures

          for on demand taxi, there is a a kayak like product. Quxsi It’s in the App Store. It compares all the ride sharing services against cabs and tells you which is cheapest.

    2. fredwilson


      1. David Semeria

        Because I think the days may be over when apps “go viral” and spread outwards from the bottom. The game has changed: home screens are pretty stable (the “core set”) and the low-hanging fruit has more or less been picked.I’ve read in a few places that startups, as a group, are among Google and FB’s biggest ad buyers. In other words it’s increasingly hard to get noticed.The big players will leverage their financial resources and prime real estate to introduce new features and enter new markets. Startups will find it increasingly hard to compete.So that’s what I meant by “upside down” — leading from the bottom will be replaced by leading from the top….

        1. fredwilson

          I totally agree and have been saying that for a while now here and elsewhere. It’s not really a new trend. It’s been that way for a while

          1. David Semeria

            Sure, but my point is that if innovation really is increasingly controlled from top, it makes the questions you asked in your post somewhat academic….

          2. SubstrateUndertow

            These are early days!Maybe its a little early to settle on the “topdown network topography” as a generic network-effect winner ?As we transition from anolog to digital culture it will take time to absorb/visualize/evaluate the palette of available network topographies and their associative attributal opportunities to solve new, as yet undiscovered, “social-topography” jobs to be done.We have just begun imagining up the new jobs-to-be-done in an abstracted net-work-effect exchange space. Not all those new jobs-to-be-done will resolve around a single net-work topographical type.We do not yet have an atomic table of net-work-effect variances and their characteristic conjugations as those memes have not yet begun to jell.The concept of federates service seems somehow pivotal here to me. I still think that the potential of things like Google-Wave have been massively underestimated.Just for the fun of conspiracy theory mongering I suspect Google killed Wave because they realized how it could decentralize their net-work-effect control πŸ™‚

          3. David Semeria

            The game has changed because the market has changed. Not so long ago being the talk-of-the-town in Harvard dorms or at SXSW was a great first step to success — not any more.The market is now mainstream. For every tech-savvy blogger or Node.js programmer, there are 10,000 FB users who are algorithmically spoon fed information by the Platform (cue Matrix references…)Basically, it’s all about control of our attention.

          4. SubstrateUndertow

            Sure it’s an attention economy but as the incumbent network-effect tool providers reach their specific network-effect peak-attention limits they will spawn new reorganized/untapped social-workflows that have not yet been visualized/imagined as per discovery/necessity dictates.As per usual new work tools will first automate the present workflows(topdown/centralized) then soon spawn new workflow possibilities to fully exploit/discover the potential of these new nonlinear/organic tool extensions. One always starts by doing what one already know/visualizes!It is the reusable-mojo at the heart of the organic-dynamic, at the heart of mutually adaptive living systems, that we have yet to collectively/intuitively grok as a coherent organizational reframing of all human culture, culture as an extended fractal instantiation of the biological organizational substrate.New narratives, metaphors and lexical memes will inevitable emerge to empower mass-culture’s indwelling of organic-process-literacy.I see no reason for that pattern not to play out here especially given the almost unlimited diversity of abstracted, algorithmically networked, organizational possibilities.

          5. sigmaalgebra

            Need something those 10,000 users will like better than FB, and JS won’t have much to do with it.

          6. sigmaalgebra

            > innovation really is increasingly controlled from topSo far, only if everyone else is just silly. We know what real innovation looks like from US national security, Bell Labs, and more. In comparison, what Silicon Valley is doing is like two naughty pre-teen boys out back picking the heads off matches trying to make fireworks.In particular, for the theme of the goals of AI in this thread today, the companies at the top are not the ones to bet on.

        2. SubstrateUndertow

          And with sites like Facebook trying to pull everything to the inside of their network you can add “inside out”.

    3. JamesHRH

      If you and Fred are attempting (as the Blockchain is) to turn a 3 component transactional equation into a 2 component one, it won’t happen.Web sites and apps can be displaced by some other intermediary but not by direct access. People’s minds work on people, place or thing. People need a thing to connect them to a person or a place.Right now, those things are sites and apps.

    4. sigmaalgebra

      Here are some “stable” things”:Horses, fire places for cooking and heating, plaster walls, oak floors, flat head auto engines, auto engine solid cam followers, auto carburetors, auto distributors and points, radio, TV, Loran for navigation, coal for heating, sailing ships, steam ships powered by coal, steam engines to drive factories, steam engines for railroads, ice houses, ocean liners, passenger trains, bias ply auto tires, steam irons, watches with main springs, phonograph records, house paint based on lead and boiled linseed oil, wood siding for houses, oil lamps for lighting, telephone dialing, clocks with pendulums, chronometers for determining latitude for navigation, silver for table settings, carbon steel for kitchen knives, tin lined copper cooking pans, soap from animal fat and lie for washing clothes, dress patterns and sewing machines, iron pipes for plumbing, solid lumber for floor joists, many traditional hand tools, magnetic tape, most books, liquid ink and fountain pens, mechanical adding and calculating machines, movie theaters, tables of mathematical special functions, slide rules, vacuum tubes, photographic film, newspapers, magazines, wooden boats, scale in auto radiators, leech bleeding, typewriters, short hand, impact printers, 700 pound, 300 MB disk drives, reference librarians, blue jeans.And of those, what’s left? Sure blue jeans.Lesson: There is progress, and a lot of things change.

  7. pointsnfigures

    Everyone setting up their own silos. Silos with network effects are hard to topple. Silos with network effects and regulatory borders are almost impossible to take down. If Uber can start to work hand in hand with government regulators to rewrite regulations, they will be impossible to stop. Banks, Insurance, Telcos and other big companies do this all the time.

    1. Joe Cardillo

      True, but does that rigidity also eventually stifle innovation? I think there’s a tension between co’s that underwrite the way things get made vs. make the things.Semi-related: trying to do both, which billion dollar co’s seem to eventually end up doing, tends to create or expose weakness in product.

    2. SubstrateUndertow

      That is the glass half empty perspective ?Maybe there are actually some social “jobs-to-be-done” where the Silo becomes counter productive. As networked social/economic/commercial interdependencies evolve many of those Silos may fine themselves scrambling to become more co-operative organs within a much more organic social-necessity .The evolution of complexity including social-complexity always come with increased synchronicity/interdependencies. Silos may serve well for physical/timely products like Uber, Amazon, Yelp, Maps, AirB&B, but for more abstracted services like Banks, Insurance, Political/Social/Economic organization those Silos come up against an immovable organic reality ?

  8. Jim Kessler

    One may say that competition is good and yet the goal is to invest in the company that dominates its market and almost completely eliminates competition.Investing in a company that is in a highly competitive market is bad. Investing in a company that will dominate its market and destroy competition is good.

  9. Kent Karlsen

    Will the future market winners use bundle or unbundle as the strategy? Example Unbundle: Niche search wins over Google on one search category. Bundle: A new technology threatens both search and ecommerce solutions and outcompete both markets. What is the most likely strategy that will get funding and win the market in the future?

    1. obarthelemy

      Can you have your cake and eat it ? FB have an unbundled Messenger, but could bundle it back into main FB while still keeping the standalone client. Modularity as Consumer version of embrace (a new type of service), extend (it by linking it up with your pre-existing offerings), extinguish (the one-trick-poney competition) ?

  10. Simone

    Q1. ‘open data platform, in which users ultimately control their data’ VS networks.I am afraid giving away our data is how we pay for being allowed to access the wonders of internet.Who will keep the lights on if we stop ‘paying’ and why? to clarify the question – what would be the monetary incentive to maintain the infrastructure of an open platform and who will want to bother doing this?Q2. ‘Will the countries with the most sophisticated tech startup communities end up winning the global economic race’?easy question, my answer is – Yes πŸ™‚

  11. obarthelemy

    Is it *that* true though, and that permanent ?You’ve picked 4 clear winners in their categories (I’d add World of Warcraft for MMO), but I think there’s quite a bit of confirmation bias. Also I’d argue LinkedIn is providing a challenge to FB.Other categories don’t have overwhelming winners: Messaging, Editorial Content (even if you break that one down by age, audience, even original vs repackager: TMZ, Digg, … don’t dominate), Streaming (whether legal or illegal, music or video), Porn (I’m unsure: lots of sites but same corp behind them all ?), Classifieds is very local.Network effects do provide some lock-in, but still, all services are one free app install away… I’m curious as to which of the unicorns will stumble first, I bet one will within 5 years.

    1. Simone


  12. Simone

    Re networks – we are gregarious, I don’t think networks will ever disappear, I guess the need for the network led to internet in the first place.People create the technology so as creators, they make technology after their own appearance, the more technology is like us, the more successful the adoption.Networks may transform over time, but can’t see them disappearing, they seem to be the purpose/a true reflection of us, not a transitory phase

  13. William Mougayar

    Been thinking about this as well, and recently wrote that underlying the new networks will be new (open, decentralized) protocols that are emerging, and it’s possible that whoever owns these new protocols would have a good chance of owning the new networks on top of these new protocols. Network effects will still prevail, but the methods of value generation and realization will change. With decentralization, users have to do a bit more work themselves, more participation, more sharing, etc., but the networked-value creation will re-start at the protocols levels. At least, that’s my best assumption so…

    1. Tom Labus

      Crossing the Chasm Part?

    2. Joe Cardillo

      I’m not sure I agree, but maybe I’m getting hung up on semantics.Fred mentions the possibility of the end of network effects, but that seems to suggest that they are something that a company can control. I don’t think that’s true of either protocols or networks. In my experience the best protocols are flexible enough to eventually generate their own replacement, and same goes for network effects.How much that can be owned, I don’t know. The pressures of public mkts, in particular, seem to restrict it. I’m not a VC and don’t know much about huge market cap co’s, but there’s something hard to digest there.

      1. Girish Mehta

        “The pressures of public markets in particular seem to restrict it…”Can you please elaborate…Thanks.

        1. Joe Cardillo

          A lot of early / fast network effects depend on non-linear growth, that is, you find a growth loop and exploit that. Some of those become really strong and turn into longer term strategy. Those are what public markets (as far as I can tell) are interested in because they contain predictable revenue over a longer time period.Many large co’s have an “innovation arm” but arguably they are more about vanity and absorbing potential competitors than actually innovating. Maybe I’m missing something?

          1. Girish Mehta

            Ok, Thanks for elaborating…think i understand what you meant a little better. I am struggling with understanding aspects of Fred’s post today…

          2. sigmaalgebra

            > “innovation arm”Middle management is there to do their assigned job, and in that job the results are supposed to be low risk. So any case of failure is a real black mark. Since innovation might fail, big company middle management won’t pursue it.Really, only a CEO can sponsor innovation.If a CEO has an R&D division, then usually the reasons are a patent shop, publicity (“Wait until you see all the great things on the way out of research” — that was just total publicity BS), and maybe a technology SWAT team when if ever needed.There are some good models for innovation in large companies, but nearly no CEO wants to pursue them.Really, basically, it’s just human nature to avoid backing something they don’t understand. So, people with power won’t and people without power can’t.

  14. matthewmclean

    I disagree Fred. There is something besides network effect at play with Google, Amazon, Uber and every other brand that is first into the consumers head with a position. “Positioning” is a marketing term that predates the internet. I first read about here over forty years ago.

    1. William Mougayar

      I’ve been a student and practioner of Positioning for about 30 years, but let’s not confuse cause and effect.In this context, you are referring to the brand position these companies occupy in our minds, and that’s a powerful factor in keeping loyalties very high. But that’s an outcome of their market dominance, and even the best brands can fall out of favor or they can get attacked and weakened anytime, especially with new generations of technologies and new generations of users.

      1. matthewmclean

        I disagree. Network effect is the result, not the cause of a brand dominating a category. All the brands mention were first to market in their category. They invented the category. Before Google, search engines were just an index of pages. Then Google invented search with page rank based on the utility of the page. They invented the category. Now all search is based on page rank.Paypal had won the position of “on-line payments” before ebay tried to introduce a competing product. Using the entire force of their network and capital they tried to kill Paypal. They failed because every idiot knows you use paypal for online purchases,Regardless of which came first – the chicken or the egg – the bigger point is that the phenomenon Fred refers to has been in play many years and really has nothing to do with the internet. The internet just makes it happen faster.Correction – Although the authors were writing about positioning in the 70’s, the book I linked to earlier was first published in 1981. I first read it in 1989.

        1. William Mougayar

          Continued growth via network effects will be accelerated when the brand is strong, but you need to have a strong brand first, which starts if you are successful with a community around your network effects. That said, not all network effect-based companies have a strong brand, but that’s another subject.

          1. matthewmclean

            Let me put it this way. Jeff Bezos started with exactly one customer. Quickly he had 1000 customers and then 10,000. That process had nothing to do with a network effect. It was the result of building a position for his brand. The eventual network effect came after he won the position. This is why little companies can compete with big ones. Thanks for the exchange William. I enjoyed it.

      2. obarthelemy

        Inertia vs quality… a large quality gap can overcome inertia ?I’m also wondering about generational phenomena (with an Internet “generation” lasting 5 yrs, not 30): should services evolve with their core/original audience, or should they stay in place and hope older users will stick with them out of habit while younger/newer users still come in ?Artists, which I consider the quintessence of a pure brand (esp. musicians: no content/service, really, just different types of noise conjuring up self-generated feelings) mostly follow their audiences after the initial harvesting.

        1. William Mougayar

          Yes, the generational user habits change, e.g. when you see new products like Kik and Snapchat being adopted by younger users, and misunderstood by older.

  15. LaVonne Reimer

    “We think it is possible that an open data platform, in which users ultimately control their data and the networks they choose to participate in, could be the thing that undoes this pattern of winner takes most.” That alone makes this a huge post for a Sunday morning! I’m really a beginner at trying to understand Blockchain but have thought a lot about decentralized protocols and systems. I see them upending our notions of ownership. I would modify what William is suggesting below. I may produce the decentralized protocol for credit analysis that everyone embraces and they may prefer the network I enable on top of it but that doesn’t mean I “own” credit. It means I’ve made it possible for every participant in commercial relationships to own credit. Winning the market to me just means I set a new standard that dramatically changed the way people behave in what used to be a black box system. Doesn’t mean all the value and/or money flows through my deal. Or another way of saying it is that this technology could foster communities who feel and behave like members of a co-op rather than customers or cogs in a corporate machine. All of this of course depends on entrepreneurs who run with decentralization as a complete philosophy rather than exploiting an aspect of it to further their otherwise game over mindset.

    1. William Mougayar

      Correct that the owning part is questionable, but at least you have a chance to provide services on top and be the majority player. But in some other cases, the protocol and the solution could be more tightly integrated, therefore benefiting its creators.

      1. LaVonne Reimer

        But those are the open questions for investors right? What’s the asset? People? Technology? I would love to see an investor (like you!) or group of investors conducting or stepping up to fund research into the business (and thus capitalization) model for “true” decentralized plays. Feels like this post is an invitation to have exactly that kind of conversation. BTW, this all feels a little deja vu. Over ten years ago, the biggest vendors in the Linux community asked me to develop a plan for an incubator that would bring many more small businesses into the Linux ecosystem. Business model questions were at the heart of this just as I believe they are with Blockchain and other similarly decentralized plays. We figured there would be some scalable startups in the mix but many would not be like that. All would play together because in doing so we’d more deeply enrich the community’s impact on innovation. But, back to the subject. Like the world of open source, I hope we don’t end up with an either its a non-profit foundation (e.g., Stellar) or scalable startups which I feel will be pushed to either owning the technology/infrastructure/protocol or owning the community of users.

    2. sigmaalgebra

      A big theme to break down the concentration at the top is very high personalization. The Internet enables enormous personalization, but it is taking a while for the exploitation.E.g., cooking: What is needed is instruction in cooking. But traditionally the best there was were cookbooks published by book publishers also more interested in drama, entertainment, etc. and not interested in instruction. So, the cookbooks were flops at being instructional. Then there were TV shows on cooking — again, entertainment, not instruction. On the Internet, mostly looking for clicks — don’t try to eat that stuff! But eventually there will be some good instructional material on cooking on the Internet, and then cooking can be much more personalized, e.g., with much more variety.E.g., for years I tried to make beef stew. I followed Child, Pepin, Escoffier, Time-Life, etc, and every trial was a disaster. I wasted time, effort, money, calories. A disaster. Eventually I found enough on food chemistry to learn that need a thermometer. E.g., simmering will NOT work. And learned that with the thermometer, mostly cook stews, especially beef stews, at about 160 F and essentially never let it get over 180 F. For some details about proteins and more in essentially food chemistry, that’s the main secret.And, never but never did I ever read a beef stew recipe that anything about 160 F or 180 F. Why? Because all the writing was for entertainment, a vicarious, escapist, emotional experience of imagining was eating good beef stew and under no circumstances actually giving solid instruction on how to make beef stew. And the situation is essentially the same for everything in cooking in books, TV, and the Internet — apparently on pain of death, no one wants to be instructional.That sad situation will change.Then with a lot of good instructional material, more people can do good cooking with much greater variety, and there will be a lot more in personalization.

  16. Brandon Kessler

    I get how the mobile phone contact list and global privacy concerns suggest a new consumer freedom which will matter to businesses.But I think network effects and the lock-in derived from them are different but related things. The network effects speak to the value users get when additional users show up on a service; this is content that users want and are happy to get, not content they’re forced to see by bad guys. If people *are* the actual content on a service, then I don’t see how this fundamental lever goes away any more than Amazon’s ability to pass on their savings to their customers goes away. So I don’t see an existential threat to network effects any more than I see it with economies of scale.But lock-in has never lasted forever (e.g. MySpace, Microsoft), especially if you don’t keep users happy. Facebook’s intense paranoia and willingness to bet the company on buying new entrants seems to be the new norm required to stay on top. What I think this all means is, these forces will still exist, but the pace of change will pick up noticeably and already is. I think it’s good for the world and investors.

  17. Chris Phenner

    I’m getting whiplash amongst posts that point to ‘Basic Income’ and ‘Winner Take Most’ (on AVC on sequential days this weekend), while seeing Forbes’ first list of top-earning YouTube stars and inequality atop the 2016 presidential campaign…Rad Idea? Invest in marketplaces that have more people participate more equally, and establish that as a strategic advantage in-and-of-itself — call it ‘Contribtors Make Most.’Imagine Etsy’s most successful sellers participating in its enterprise value, and that wealth created funded some social good(s).What’s duplicitous: All the talk on this site about the seemingly benign (eg ‘kids coding’ so closely alongside the ravenously unequal (‘network effects’). What if these values were aligned?Recall that for every Kickstarter-like call for benign funding for social good on this site, hundreds of millions of dollars in IRR is being created in pursuit of oligarch-style wealth.Reconcile that if y’all need a big topic to chew on.

    1. fredwilson

      Kickstarter, Etsy, Shapeways, SoundCloud, Wattpad. We’ve got those bets on our portfolio

      1. Chris Phenner

        I read AVC multiple times weekly, and I know much of USV’s portfolio (almost from memory) — my post is not meant to suggest USV should ‘make more investments’ like Etsy, Shapeways, SC and the others — I have been watching USV for several years.What I am trying to suggest — and I may not be doing the most articulate job — is that there is an inherent irony (or dissonance) in the posts on AVC and in the remarks of VCs like Sam Altman and lots of others.Here’s another way to describe the irony/dissonance:1. We invest *our* LPs’ dollars for network effect / out-sized returns. **2. We wish for fairer distribution of education and economic opportunity.** In case I’m not getting through: The investment thesis of USV that targets ‘large, engaged networks’ is predicated on creating out-sized returns and wealth outcomes that flow disproportionately to a small number of folks.Put another way: At the end of Day One of Esty’s IPO, How did the wealth of the largest shareolder of Etsy (preferred?) stock end up relative to the wealth of the largest Etsy seller? I don’t know the answer to the question I just asked, but my sense is that an early-stage investor ended up FAR better off.Where I see hope: Your post about that new corporate structure for Kickstarter was encouraging — and fundamentally different-sounding than the ‘out-sized returns’ language related to USV’s investment thesis. At least it read differently to me than what I read in AVC’s early days re ‘Thesis’ posts.I know USV makes progressive, network-related bets, and that it has for years. What I am posting about is much more fundamental than bets that improve the IRR and coffers of USV and its LPs in out-sized ways.My post is wondering aloud about how ‘enterprise value’ (or IRR) is distributed in a manner that it creates more wealth (or stability) for a much, much greater number of people than what USV is ‘betting on’ today.I hope that clarifies. It’s neat to be able to think aloud, this big.

  18. Matt Hardy

    As I see it the key distinction here is around language, not country/region.English is already the most universal language in the world and new translation technology will only exacerbate that fact in the analog -> digital -> hyper digital transition. One way to look at this is it will concentrate access to markets even further, having the opposite impact of your hypothesis, Fred.I’m picturing an advanced version of google translate that instanteously delivers every message to a person in their native tongue regardless of what language it was input in. This further concentration of influence would only be short term as eventually the value from all inputs would be sussed out meritocraticallyy regardless of source.The disruption during that interim period of further concentration, however, could end up having lasting influence.

    1. obarthelemy

      Might depend on language, but country/region certainly matters. I’m French, and can attest French people use the Frenchie-French part of the Internet, not the Belgian-French, Canadian-French, Swiss-French, Congo-French…That’s obvious for services with a real-world impact (shopping…) but also true for purely self-contained websites/services (text, videos, news, blogs, messaging…) because of small language differences (the fr-fr for day-care means orphanage in fr-ca, it’s fun the first few times you announce you’re taking your nephews to the orphanage ^^) and not-that-small cultural differences.

  19. Bipper Media

    It sounds like the question you are posing here is “are there any secrets left to discover?” on the internet and mobile? (as posed by Peter Thiel in “Zero To One”). For those not willing to do the work, ask the questions, and start small, then the answer is “no”.

  20. iggyfanlo

    “We think it is possible that an open data platform, in which users ultimately control their data and the networks they choose to participate in, could be the thing that undoes this pattern of winner takes most.”I don’t think that this condition necessarily changes things, but I DO believe that the “amount” of producer surplus that currently exists for the winners (Uber, Google, Amazon, etc) gets reduced dramatically… they may still win, but margins will be squeezed and/or directly transferred to users

  21. greyenlightenment

    winner take all US economy, winner take all web 2.0 economy

  22. ZekeV

    I think you and your partners are right, or at least I want to believe you are. Eventually users will be in control of their own data, and whatever protocols emerge that accomplish this, will win over the monopolies that silo user data. You could also view this as an open network effect that benefits users, and creates a defensible barrier to competition from aspiring monopolists.From a venture investing standpoint, this breaks the model of buying early-stage equity in future monopolies. How do you do that if the monopoly-magnitude gains are falling to the users rather than a company? In micro terms, how do you capture a portion of the consumer surplus?Bitcoin provides a hint. To the extent that open network protocols embed scarce digital tokens, you can still go after venture returns by purchasing (or otherwise acquiring) tokens in protocols that you think will come to dominate the landscape in this new environment. You might still want to fund companies that provide value to the network, but you would not necessarily look for those companies to produce the big venture-funded exits of early web companies. You would fund them almost altruistically, and look to the digital assets for upside.

  23. Shaun Dakin

    Hasn’t this been the case in every category for decades? Consolidation is the general direction of every single category. For example, when I worked in the office of the chairman at Fannie Mae 2001 – 2004 we studied the massive banking and mortgage consolidation over decades.

  24. howardlindzon

    Welcome home, this is awesome, especially on the eve of Stocktoberfest where we are thinking about this from a valuation standpoint, law of large numbers and who might be the first TRILLION dollar company. Wish you could be here to discuss, But I will be sure to incorporate this post into some discussions on markets with the traders and entrepreneurs in the room.

  25. howardlindzon

    Interesting that Exxon, Phillip Morris and Apple not mentioned here…maybe no network effects but maybe the first trillion $ company uses others networks the best

  26. jseliger

    Will that always be the case? Will the countries with the most sophisticated tech startup communities end up winning the global economic race as we transition from an analog to a digital world in which the winners take most of the market?It might be. In terms of human languages, English appears to have “won” the global marketplace for the foreseeable future: though Chinese has more speakers, English is the world’s most taught second language and has a strong presence in India. Barring nuclear war or something similar it’s hard to see another language supplanting it.If one sees it in language itself, it makes sense to see it in other domains too.

    1. Girish Mehta

      India has the world’s second largest population of English speakers after the US, but that is because, in India, small percentages are large numbers. ~ 88% of the population is non-English speaking.Earlier this year India also overtook the US as the second largest market of total Internet users in the world. There are now more Internet users in India than the population of the US.The number of Internet users in India is already well greater than the English speaking population in the country. India will get to 500 million Internet users by 2017-2018 and simple math says that the growth will come predominantly from non-English users..

  27. charlieok

    β€œWe think it is possible that an open data platform, in which users ultimately control their data and the networks they choose to participate in, could be the thing that undoes this pattern of winner takes most.”There is a community pursuing this idea (and there has been in some form for years, maybe for as long as we’ve had centralized behemoths).In fact, there is a Redecentralize conference taking place at Thoughtworks in London this weekend:…They are posting a lot of stuff (somewhat ironically) using the #redecentralize hashtag on twitter:…

  28. Joe Cardillo

    Seems like a good moment to re-up this interview w/Marc Andreessen, in particular the part where he talks about disintermediation:

  29. Jon Smirl

    Probably the most valuable thing in the world is the Internet itself. And ownership of it is completely decentralized to thousands of companies.One thing that does seem to be changing is the role switching costs play in the maintenance of winner take most networks. For example Google search is a winner take most and the switching costs away from it for consumers is essentially zero. Yet people stick with Google. Ask AltaVista if winner take most protected them. Or MySpace had any protection from Facebook.On the other hand Windows is a monopoly with extremely high switching costs to escape. It took the creation of the smart phone and tablets (platform shift) to escape the clutches of that winner take most market.If you want another good topic to discuss, consider the impact of apps and apps stores with web sites being packaged as applications on the two major platforms (android/IOS) versus if app platforms did not exist and everything instead were HTML5 based. There is no technical reason every app in those app stores couldn’t have been written in some flavor of HTML5. But then we wouldn’t have platform lock-in which benefits Apple/Google with their 30% cuts.Job’s greatest achievement was not the iPhone, it was taking control of the app stores away from the cell vendors and then turning it into a major platform.

    1. sigmaalgebra


  30. dave schmierer

    I’m surprised that Europe has been so willing to allow US startups to capture their economic value. AirBnB and Uber are sucking commissions from European markets and sending them back to San Fran. Not surprising to see some pushback as this grows.

  31. Matt A. Myers

    It’s governance in the end that will help dictate who will win in the end – so long as the “free market” isn’t impacted – which it has shown to be influenced by things like U.S. government spy programs gathering and telling U.S. company secrets of foreign competitors.An open data structure has the potential to disrupt all – a layer for everything else to evolve on top of – however how that is funded is the issue.You need every piece of Facebook and other mainstream platforms to be open source, protection free, and to be able to be forked in pieces – so then someone can duplicate everything quickly, have access to the data being easily displayed, and so then the defensibility of the network effect becomes moot.Will any investor actually invest in this though? Not without a return, of course not.Would a society invest in this? Not with proof of concept, of course not.Is there a middle ground? I think so.

  32. sigmaalgebra

    Network Effects 101(1) Shopping Malls.At the best shopping mall in town, all the best stores are there because all the best customer are there because all the best stores are there.(2) Country Clubs.At the best country club in town, all the best people in town want to join because all the other best people in town are members.(3) Expensive Cars.As no doubt Alfred Sloan noticed, the people with some extra money who want to show off buy Cadillacs because a lot of other people with something to show off bought Cadillacs. They don’t want to buy a Chevy because, ugh, a lot of people you wouldn’t want your daughter to date drive Chevys.(4) Luxury Brands.Everyone wants to give a gift from Tiffany’s because all the best people give gifts from Tiffany’s.Similarly for lots of high end, luxury brand names.That is, for Tiffany’s, mostly we are talking brand name, not network effect.Ah, any other big jewelry store, Audrey Hepburn, a movie, Reese Witherspoon and another movie, a location on Fifth Avenue, and, now, one of Trump’s daughters could do the same!(5) The Internet.Everyone wants to connect to the Internet because nearly all the content is on the Internet because nearly everyone is connected to the Internet.Then, sure:(6) Facebook.Mary is on Facebook because all her friends are.Of course, there is the ad, Bum 1: Are you on Wo-Wo?Bum 2: Everyone’s on Wo-Wo.Overhearing this conversation,Manager: We’re going all in on Wo-Wo.Soon,Bum 1: Are you still on Wo-Wo?Bum 2: Naw, man. I’m off Wo-Wo. Of course the people who developed this ad expect that nearly everyone will understand it right away. That is, there are fads, and some grow quickly and also die quickly.GoogleThey are a lot better? IMHO Nope: Easily enough Bing has been able to be quite comparable. Apparently also Yahoo and a service in China and another one in Russia.So, Google is a little better. They got a lot of users when they were a lot better. Then they were awash in cash and able to hire a good team of software developers, polish their service in a lot of details, and stay a little better. IMHO, still, at times Bing is better.Google, has some economies of scale? Likely.Google has a network effect? IMHO not in any usual sense. E.g., that Mary uses Google is no great reason for Joe to, whether Mary and Joe know, or want to know, each other or not.Amazon.They have a guy, Jeff Bezos. He’s a smart cookie.Wal-Mart, from their (A) relationships with vendors, (B) buyers, (C) warehouses, (D) trucks, and (E) many stores, has had a great opportunity to beat Amazon.But when I want a good supply of 100-200-300 W mogul bulbs for my office floor lamp and am tired of working to find them at Wal-Mart or the local hardware stores, then I just order a box of six from Amazon. Also when I was snowed in and needed some tire chains to get out, I ordered, Amazon delivered, and I got out. When I want a particular variety of Campbell’s soup, am tired of struggling at Wal-Mart or A&P to find, say, a dozen cans, I get a case of 24 cans from Amazon.Here Wal-Mart blew it. They just sat there, rich, dumb, and lazy in Bentonville, and watched Bezos grab much of their business.IMHO, Amazon is from Bezos and Wal-Mart being asleep at the switch, not from any network effect.But, it’s possible to beat Amazon. E.g., as I shop for my next PC, I can get some okay to good advice from the sales staff at TigerDirect. Or, I can buy from Sam’s Club who will accept returns for essentially any reason. So, if I buy some PC part and my motherboard BIOS’s is not up to the challenge, then Sam’s will accept a return.Recently I tested Sam’s return policy: I wanted some lunch sandwiches of sliced, roasted turkey breast with lettuce and tomato. So, I bought some chunk of cooked turkey breast.Ugh! Unwrapped and sliced, it didn’t look like anything that had ever been alive. Unless I sliced it really thin, it made me nauseous.So, keep it an feed it to my kitty cat? Nope: My kitty is smarter than that; wouldn’t touch the stuff.Okay, put it on the back porch for the crows; they will eat anything. But they were also smarter than that: I had little cubes of it sitting out there for days before I gave up. Finally I just returned the whole mess and got a full refund.UberRide sharing is strongly a geographically local business. That Uber is successful in San Francisco is no great reason they should be successful in Boston, NYC, Paris, Rome, or Warsaw.I don’t know why Uber is so successful — that is, if they really are, which with their burn rate maybe they are not.IMHO, there’s no great network effect in the ride sharing business. E.g., if Lyft works hard on Peoria, they stand to win in Peoria. Then, using their successful brand name in Peoria, they can expand to neighboring cities.A network effect is terrific when it works. But there tends to be startup cost getting to critical mass. In that case, start in a niche, an entering wedge, and expand from there.Let’s learn from the past 100 years of aviation: Early on the big secret was linen and dope.Then the big deal was a V-12 engine with a gear drive. Next, a big radial engine with direct drive. Next, understand Reynold’s number and how to scale the data from a wind tunnel, and, thus, get rid of biplanes and have much better lift to drag ratio with monoplanes. Learn to make frames from 4130 chromoly steel. Learn how to work with aluminum and flush rivets. Next, super chargers. Next, Carnot’s cycle, high compression, and aviation gas from the addition of tetraethyl lead. Then a two row radial. Then a four row radial, with some use of magnesium. Then jet engines. Then axial flow jet engines.Then notice that the engine moves a mass of air m to a velocity v. So, put kinetic energy E = (1/2)mv^2 into the air and get out to drive the airplane momentum p = mv. So, pay E and get p. So, since in E the v gets squared, if have m larger and v smaller for the same p, then need much less E. Or for the same E, get much more p.So, have a high bypass turbofan jet engine.And on and on.Well, Moore’s law, microprocessors, computing, software, and the Internet are information technology. So, we take in available data, manipulate it, and put out data that we hope is valuable information. For this work, part of what we need to do is (A) get some good input data and (B) do some powerful manipulations of it.In that work, part (B) is the main opportunity for progress analogous to that for airplanes and to compete with network effects, brand names, etc.

  33. Ana Milicevic

    “Will the countries with the most sophisticated tech startup communities end up winning the global economic race as we transition from an analog to a digital world in which the winners take most of the market?”Yes and no – the reason we’ve all been adopting Valley technology as quickly as we have is because that’s the one place that has consistently been productizing innovation and taking it to market with strong (and predictable) capital backing. Valley companies are the first to turn into verbs (Googling, Ubering) and the first to transform into utility — the way we don’t think of who supplies our electricity all that often is the way we’re increasingly (not) thinking about who supplies our search, our connections with friends, or that car that’ll drive you home from the bar at 3am. What this is enabling in non-Valley markets however is innovation on top of this layer of utilities. The most interesting example for me lately has been how commerce evolves over Whatsapp, pushing the platform into a territory where it doesn’t necessarily reside natively in developed markets, but potentially a huge boon in developing ones. The biggest driver of localized innovation hubs will be the availability of capital to sustain and grow them, and here I don’t think we’re looking at a Valley-like model at all which is why it may be hard to spot them: we expect this to look like ‘Silicon _____’ the very same way we’re holding our breath for the next Steve Jobs to come along.

    1. Donna Brewington White

      Innovate or be disrupted seems to be the message taught by tech. Not just for industries…this goes for models and thinking as well, eh? Harder to disrupt these but desperation of those trying to break in might drive this.

      1. Ana Milicevic

        Yes – but also every iteration brings new opportunities. While Facebook & Whatsapp may be the new utility layer (like electricity) where subsequent innovation is less visible but more about optimization (e.g. better data storage, improved personalization, predictively surfacing content & relationships, etc) there’s a layer of commerce appearing organically on top. A great example of this would be Beam in Ghana ( — essentially a TaskRabbit for all types of tasks but built on top of Whatsapp, a platform people already use and have on their phones. I discovered it by accident about a week ago and can’t stop thinking about the genius of its business model.

  34. Donna Brewington White

    Will the countries with the most sophisticated tech startup communities end up winning the global economic race Will tech startup communities always be defined by geography or nationality?

    1. sigmaalgebra

      “Always”? Always is a very long time, so, no, not always. For a long time? Yes.There’s a lot more innovating to do, and as far behind as the US is from what we could be doing, others are still farther back. The US stands to stay ahead for a long time.

  35. Anu Hariharan

    The problem with the rest of the world is they are lagging behind in product innovation. Take India for example – I think the reason US companies (Uber and Amazon) are giving a tough fight to local startups is due to the lack of product innovation. China on the other hand does seem to innovate faster and in some cases have even been pioneers in new product innovation (e.g., WeChat) But companies are learning with time and now recruiting silicon valley talent by matching their US salaries to work in startups in Bangalore to focus on product innovation.

  36. creative group

    Do consumers really choose what platform wins the sector or segment of any particular market category?Do the VC king makers really chose who ultimately wins in each sector?The billions used to push scale appears to be the real key to the success of getting the apps, sites and what the VC groups think is the next big thing in the hands of consumers to make that choice.

  37. Terry J Leach

    No the countries with the most sophisticated tech startup communities will not win the digital transformation race. I believe your thesis is correct in that the blockchain will be the open data platform. The Blockstack for a blockchain based platform is looking good enough for my startup to build a non-financial based platform the enterprise market which we are very familiar. The installation phase is near completion. I think the transition will play out at the edge of the enterprise in “green field areas” that the current winners do not play. Once a problem of sufficient magnitude is solved by a open data platform with a great user experience using the blockchain then the development model will copied by others and spread widely. The blockchain based enterprise will not own customer data, they will have to provide great service and have a low take rate and will not be based in any particular country.Bootstrapping a blockchain based open data network requires a “come for the tool, but stay for the network” approach ala Delicious. We will witness the transition from the 20th century firm to protocol based network that governs the interaction of people on the network. At some point the time between the installation and deployment phase as described Carolta Perez will not exist.

  38. Bernard Desarnauts

    Where is Apple in this analysis??? yes 10-20% of the volume but 95% of the value/margin???

  39. Douglas Crets

    The answer to your question broadly is no.The most sophisticated tech communities won’t win. For two reasons: One is that there will be too many, and too many different types of tech communities that the next wave of consolidation will not really happen.And two, Because of culture and the way that Internet is being used that both predates major tech players in North America and Europe and also has been built up in ways that avoids them or doesn’t need them.I wrote a little about my visits to China and the way I see tech being used there.… /notes-from-behind-china-s-great-firewall-the-people-s-content-is-the-people-15b47c64b6b6China almost in a completely opposite manner looks at tech in ways that the North American audience hasn’t considered, or can’t consider because of regulation.

  40. kidmercury

    i agree the end is an open data platform, though i think it will take the form of a federated approach. this is one of the problems with bitcoin, it is so anarchic that even federation becomes impossible.the US has the benefit of its capital surplus, though i find it dubious that this is a sustainable advantage. if/when the capital advantage is lost, we will have to look at which nation-state or collection of city-states really has the tech skills needed. it might still be the USA or california, though i think south korea/seoul will be a serious contender.

  41. Jesal Sangani

    2 cents: Winner will still own majority of the market due to inherent advantages. I believe there will be multiple outcomes: 1)Market Protection/Regulation (Developing World), 2)Hyper-focus “Niche” in serving under-served customers, which can disrupt(Multiple winner for each geography), 3)Too big to move: Current winner becomes too big and may miss next S curve, and new core technology may arise from different part of the world.

  42. george

    There are a number of juicy soundbites above, maybe I’ll just touch on platforms (networks). I think all of the companies mentioned above have proven out one thing, data creates value and it has transformative capabilities. However, mobility and smartphones have created the sphere and without it, scale, utility and experience would never have hit this level of emotion or reward.In my view, the device (iPhone, Galaxy, Nexus and etc…) is the the closest thing to what you describe as “the user controlling the data and the network they choose to participate in.” Seems like Apple understands this the sphere pretty well and so have their users as they control more and more choices.

    1. Douglas Crets

      I really like the direction of this thinking.

  43. julianranger

    Fred you state: “We think it is possible that an open data platform, in which users ultimately control their data and the networks they choose to participate in, could be the thing that undoes this pattern of winner takes most.”You and I had this debate a couple of years ago and I know I failed to convince you then, but you’ve come to the point that I made then that “users ultimately control their data”. Now the question is how? – you state the blockchain as one solution. My view is that it is by the individual themselves. If companies own and control their own data (think SAP et al) and the military do the same, why shouldn’t we as individuals be like mini-businesses and have the same data architecture? – IF it is simple to do. We have been working this process at and it does solve a lot of the issues with personal data for both sides of the equation, individuals and businesses. If you have 13 minutes (I know you’d like 2 minutes, but like a good interview sometimes longer is better) please look at

  44. csertoglu

    This is a point we discuss at our firm frequently. We think how these markets evolve will impact their economics, and in the case of marketplaces with network effects, it will affect the take rate of the marketplace owner. In early stages, we see the early leader, looking like a winner-takes-all type winner, will enjoy great take rates (like Uber seems to be these days). If it’s not a winner-takes-all market at the end of the day, the take rates will likely mature at lower levels.

  45. christopolis

    much of the internet centralization is due to macro economic policy. centralize power and you centralize power. The last 20 years has seen a vast centralization of power in the US federal government. A reversion to the mean is inevitable. Once this happens you will see diversification and localization become a much stronger force. Tons more innovation, craft search engines, marketplaces etc. Follow the Beer industry prior to prohibition through today for the way it would play out.

  46. Alex Bugeja

    The refinement I would add to this is that the winner takes most of the market at one time, until the next best thing comes along. Google took search from Altavista, which was considered the winner at the time.This is less likely to happen when there are strong network effects as the article mentions. Facebook is obviously a good example. Search is less clearly an example of network effects to me – in theory replacing Google is as easy as clicking over to Bing.

  47. Joel Natividad

    The open data platform that has enabled all this innovation is the Web. In my mind, its next iteration, the Semantic Web, is the nextgen Open platform that nobody and everybody owns, will evolve to enable “Winner Us All.”Because, unlike the current Web where network effects result in the concentration of Data/Power in Hubs, the SemWeb, if its potential is fully realized, is a meta-platform that distributes connections, rather than concentrating them.It has fallen out of favor of late, but I’m hopeful that it will rise again, once the enabling technologies are in place, following the same trajectory of A.I., when it was not taken seriously in the 80s/90s.I make it a point to periodically read Tim Berners-Lee’s original Scientific American 2001 article to remind myself of its potential and to keep the faith –

  48. steve cheney

    Interesting to see you think about whether this will continue. I wrote this 2 years ago on the consumer web being binary – Andy W commented in DISQUS.

  49. RadhaKrishnan Venkataramani

    I think you forgot `Recruiting`: Linkedin. Linkedin has created the recruiting platform with network effect and created a very tall gigantic barrier for anyone to enter.

  50. Jake Gold

    > “We think it is possible that an open data platform, in which users ultimately control their data and the networks they choose to participate in, could be the thing that undoes this pattern of winner takes most.”You’re spotting a trend here that’s beginning to bubble up. Every hacker I talk to these days agrees that things are about to change radically. The question is how. The blockchain is the flashiest idea, and it does solve some problems, but it’s not a practical solution to much of anything right now.The original promise of the internet was that we would all have our own domains and run cool services on them, email and web were just the beginning. But it was too expensive and too complicated to do it yourself, so the big companies saw the opportunity to lock people up on their domains and took it. We all became [email protected],,, etc. New open protocol development came to a standstill. Things that should have been open protocols, like web search, social networking, and microblogging were instead created as proprietary services.Users having independence on the internet was good idea in the 90s, but it wasn’t possible to deliver it. Society and technology needed time to catch up. Now CPU, network, and storage are super fast and super cheap. Domain prices have plummeted. SSL is becoming free. Virtualization is incredibly efficient, open source software is orders of magnitude better, there are far more internet programmers, the importance of privacy and security are more widely recognized, broadband adoptions is high, etc.When every person has a domain and cloud server they control, there will be a market for cloud software that dwarfs all existing software platforms combined. It’s coming soon.

  51. Katryna Dow

    Great post and very encouraging.The move from winner takes most to distributed wins is part of an overall shift that bridges the evolution of economics and society. The last two decades have proven the value of networks and multi-sided markets.If these models are to mature then all the participants in the value exchange need to share in the value.As personal data becomes key to unlocking personalised and on-demand services we will need mechanisms for permission and context.We have been working towards an open data platform at for the past 18 months, and beginning to incorporate blockchain components. Essentially we believe that data is an asset, and like all assets should return some value to the asset generator, enabling individuals to freely move between networks and services on their terms.Here more context in a blog post I wrote, including inspiration from your post. Huge thank you, KD…

  52. obarthelemy

    some pure lock-in, some social lock-in (network effects):one-stop shopping: not just groceries, but anything you buy weekly or monthly (cleaning supplies, toiletries…). group loyalty card: loyalty credits based on a group, not an individual (level 0: free shipping subscription)recipes AI: deduce what user is cooking, suggest all ingredients (level 2: suggest ingredients when running out, using flour-to-mapple syrup ratio for pancakes, etc)gifting/couponing within groups: let users add trial samples to relatives’/friends’ orders, with gift wrap and note. I’m always bugging my mom and sister for product/recipe ideas, and they reciprocate. Much more comfort-zone than looking up recipes/reviews.

  53. JLM

    .May not strike the chord you are looking for but “organic” everything is out there. I walked into a convenience store yesterday and they had “organic” pears — no other organic food just pears.I was next in an “ordinary” grocery store (something I don’t do too often as my purchases reflect my tastes which tend to run toward different kinds of salami) and was amazed at how much organic stuff there is. I asked the manager where they got there stuff from and ???? he didn’t know just knew it came in labelled organic.I next went to Whole Foods/Paycheck and the museum quality presentation of their non-organic fruit had me under its spell and I bought non-organic radishes in sufficient quantities to prepare for the Apocalypse.Organic is an organizing trend. [Might be a headfake also.]JLMwww.themusingsofthebigredca…

  54. falicon

    prob. would come down to pricing models (the more people who buy this, the cheaper we can offer it) or some form of tiered product offering (the more people who buy this, the more additional options or products we will offer).I think overall it’s pretty hard to do network effects with physical goods though because the value is ‘baked in’ and pretty personal/individual (you eating better doesn’t really benefit or impact my eating directly) …you can game the system a bit with schemes like I mention above but it won’t be ‘true’ network effects (and so results will vary drastically)

  55. Richard

    Best example of this is to hook your product onto an existing network. Coconut water did this well with bikram yoga. Orgnic bread, needs a great name, shape or new ingrediant (muscle bread with seeds) focus on teens and summer camps ?

  56. LE

    my tastes which tend to run toward different kinds of salamiI am in awe of the food that you can so easily still digest. I stopped being able to eat salami many years ago. I used to make salami and eggs (and cheese) but even then I used to – after cooking the salami to perfection – pat it dry to remove all of the excess fat and oils. Leaving an almost jerky like beef product all crusty and good. Haven’t done that in years. Ditto for pepperoni. And don’t even get me talking about the steak sandwiches that I had while working as an office boy during college.and the museum quality presentation of their non-organic fruit had me under its spellAnd if it had been in a Soviet era cold war supermarket [1] would you have been so captivated? According to what you are saying the answer is no.These fits nicely with my theme of food as entertainment and pleasure being the reason for many of our countrys health problems. Look at what food porn has done to us. [2][1] You remember the way supermarkets used to be with the ballast impaired fluorescent lighting, right?[2] Has it occurred to anyone how desktop publishing and computer power has led to the state we have now? It has. Back in the linotype and even pre-desktop publishing days we never had packaging and store design so nice as we do today. And advances in the label and packaging industries as well. One of the few places that printing is still doing very well.

  57. awaldstein

    You are suggesting that WF will allow for a buyers type club to exit within their stores?That WF will allow a different SRP for a different class of shoppers?Sounds like food coop concept which I buy into but I would bet wf wouldn’t.

  58. awaldstein

    Yup… that is whole foods in a nutshell.And it will change dramatically towards lower non organic in the next couple of quarters is my bet.The layoffs–their first–are a bellwether for them.