The Evolution Of The USV Thesis (and

USV is a thesis driven venture capital firm. Brad and I locked into that notion at the very start of our partnership in the summer of 2003. We decided that we would have a thesis, stick to it, and evolve it. And we have done that. We and the partners who have joined us over the years are very proud of that.

Our partner Andy has now written two blog posts outlining our thesis. He published the first one in May 2012 and he published an updated one yesterday. I encourage everyone to head over to the updated and read it.

The thing about our investment thesis it that it evolves over time. We stick to it but we evolve it. And we do it consciously. Writing it down forces us to be clear in our thinking and draw distinctions that matter. The evolution sometimes includes leaving a sector (like ad:tech which we have not invested in for almost ten years now). But mostly it involves adding new areas while maintaining others.

This chart from Andy’s post shows how we have evolved our thesis over time. This is a chart of active investments so anything we have exited or has shut down is not included in this chart.


Andy’s post explains how we have evolved to this set of investment theses over time. This is where we are right now, but if there is one thing I am certain of, it is that this is not where we will be a few years from now. The evolution will continue.

Speaking of evolution, we have also evolved our website. It started out as an online brochure in 2003 when we started raising our first fund. Then in 2005, we turned it into a blog. I think we were the first VC firm to make our homepage a blog. But we didn’t blog that frequently on so the page was not that lively. We tried to fix that with a redesign we launched in 2009. That was a lot better but we still didn’t have enough action on the page so in 2012 we started a process to reimagine our website. That led to a linkblog that we launched in 2013. Ultimately that didn’t work that well for us either. So this summer Jonathan and Nick started hacking on another idea, which is to focus the home page on the topics that are most interesting to us right now.

Our topic centric website is now live at We have a new design and color scheme. But mostly we have a new way of showcasing our thoughts and interests. I think it will work much better for us and hopefully for the entrepreneurs who show up at and want to know a bit about us. Give it a spin and follow some topics that you share an interest in with us.

#VC & Technology

Comments (Archived):

  1. Daksh

    I wonder if the chart could be depicted in a manner where evolution of the thesis was easier to comprehend. Perhaps the Y-axis could have been in terms of % or perhaps a different graph to show how each sector has evolved.An example of comprehension is that currently one is forced to look at the parallel difference in the Ad Tech line to map it to your statement of “no investments in almost 10 years”

  2. awaldstein

    That’s for my morning jolt of thought.It will be interesting to see that as your investments focus on more specific niches and enablers, rather than broader platforms, whether the time to market traction and exit is shortened.And whether at that level market dynamics you can have large exists without being the dominant market player.

  3. LIAD

    I like the pragmatic, cascading, ‘mercenary-esque’ evolution.I see it as:Be the Network—-> Support the Network———-> Undermine the Network——————> Obviate the Network

    1. awaldstein

      do the networks really need to fail for the niches and supporting apps to win?I can’t imagine a vertical like kickstarter for example being as successful without living cross the large platforms where its community lives.

      1. Girish Mehta

        Agree, I don’t think its one or the other. As Andy says towards the end of his post – the thesis is cumulative.Such a point-case might arise in the future (and USV would need to comprehend that in at that point)…but not seeing it as inevitable about the need for networks to fail for niches and supporting apps to win as a overarching theme.

        1. awaldstein

          Nicely put.I honestly think the evolution is that the idea of niches just goes away and communities themselves become more spontaneous and atomic by nature.

      2. Mark Essel

        I think its more of an erosion of competitive advantages due to owning a large network. Kinda like faxes today, they exist and are greatly used but not as ubiquitous (I take photos and email).

        1. awaldstein

          Hey Mark!I agree but just so pleased to see you.

          1. Mark Essel

            Howdy! I’ve been commenting a little more after diving back in to another startup. Always more to learn, and I’ve focused a little heavily on technical learning the past few years.

          2. awaldstein

            I still miss your blog.Coffee or a drink? Tis the season and you know my contact points.

          3. Mark Essel

            Yes. I’m downtown again will connect. Breakfast was fun last time. Coffee would be refreshing

        2. ShanaC

          🙂 Hi thereAnd that is heavily industry dependent. I actually need to keep a list of fax numbers for doctors :/

          1. Mark Essel

            Heyo!Gotcha, faxes aren’t gone by any means, but a decent fraction of folks can get by without one.

    2. Mark Essel

      Hah! Love this view good one LIAD

    3. fredwilson

      That’s clearly where we’d like to go if the market wants to go with us

      1. JamesHRH

        New site is engaging & transparent. Very now.

    4. jason wright

      what is a network’s next move to defend against this?

      1. LIAD

        Thus far. Reality.

        1. jason wright

          removing one incumbent network and replacing it with another type of network that will soon become the new incumbent. a star is born, a star dies, a new start is born…, gravity, explosion, gravity…IPFS et.c…

      2. Matt A. Myers


    5. Nick Grossman


  4. jason wright

    pivotish.the redesign works well on my mobile phone.

  5. Twain Twain

    Blockchain App Stack. Ethics of Algorithms. Machine Intelligence.Elon Musk: “I think the best defense against the misuse of AI is to empower as many people as possible to have AI. If everyone has AI powers, then there’s not any one person or a small set of individuals who can have AI superpower.”Thesis parallels here with decentralizing data à la Blockchain and putting it into hands of many — maybe so they can monetize it differently from what happens today via incumbents.The Ethics of Algorithms piece unifies two spheres of Blockchain and AI. IoT is other cornerstone.(edited for brevity).

    1. creative group

      Twain Twain:When the subject involves computer science we usually focus on what you have contributed to the topic in this space.The discouraging issue we have is that the well funded (VC’s and Wealthy blocks) groups acknowledge a technology and exploit it monetarily where the average people can’t enter or maintain the space. A prime example is bitcoin which was supported by the average techie and libertarian sector and the money pushed the majority of them out with ASIC chips that required millions to build in the first stages of major money entering bitcoin. (Antminer about $6,990)Anyone without the money and invitationto the exclusive meetups should be more than alarmed and worried.Ditigal Coin by Nathaniel Popper outlineswhat is happening not what can happen.

      1. Twain Twain

        In AI, we now have billionaire friends going up against each other’s models:(1.) 1st camp = Larry Page, Sergei Brin, all of Alphabet(2.) 2nd camp = Elon Musk, Peter Thiel, Reid Hoffmann, Jeff Besoz via AWS(3.) 3rd camp = Mark Zuckerberg, Bill Gates(4.) 4th camp = Baidu, IBM Watson, Intel.Apple is yet to show its hands.The funniest thing is that the Ethics of Algorithms and how to tool the machines to do Natural Language Processing so they understand our meanings, languages and values …CANNOT BE ACHIEVED WITHOUT FEMALE KNOWHOW.And there aren’t any female billionaires who know enough Quantum Theory, economics and code to add their 2 cents.Still … it is a female inventor who’s built the tools.

        1. Simone

          That is a beautiful thing to say, and I know there is a deeper meaning in relation to how our brains work differently (but we have both systems incorporated, regardless of gender), I never thought about it this way. Can you please share the source for your ppt, is it on slideshare? I could use it for a crowd writing project I have recently joined. Thank you (I really hope you are a guy, your comment would be even better)

          1. Twain Twain

            Ah, LOL, it’s better I’m a woman. Being a woman (and maths grad) I’m able to connect and integrate the different pieces of the puzzle in a logical and systematic way that’s different from how guys do it!The University of Pennsylvania research on neural pathways is linked to here:*…The John Hopkins and University of Sydney research on language and communication in our brains is here:*…The front <=> back propagation is from Neural Networks literature:*

          2. Simone

            Lost all hope again :)!Thank you so much for the links, they couldn’t be more timely for me.And thank you for your comments in general, I always find them informative and thoughtful.

          3. Twain Twain

            Keep hopes burning bright because … the greatest breakthroughs in tech ALL NEEDED WOMEN WORKING WELL WITH MEN TO MAKE IT HAPPEN & WORK.1830s: world’s first computer = Ada Lovelace + Charles Babbage1941: Wi-fi, Bluetooth + CDMA = Hedy Lamarr + George Antheil1944: Harvard Mark I computer = Admiral Grace Murray Hopper + Howard Aiken1959: COBOL language = Admiral Grace Murray Hopper + UNIVAC team1969: Apollo 11 moon landing = Margaret Hamilton + NASA team1971: Computer telephony switching = Dr Erna Schneider Hoover + Bell Labs team1972: Smalltalk language (later Squeak) = Adele Goldberg + Alan Kay + Xerox PARC team1974: CLU language = Barbara Liskov + MIT team1985: Spanning Tree Protocol (STP) for the Internet = Radia Perlman + Digital Equipment Corp team1990s: ATMs & distributed transaction processing = Dr. Mandy Chessell + IBM team1995: Affective Computing = Rosalind Picard + MIT team2015: Virtual Reality = Mary Lou Jepsen + Facebook/Oculus teamPlus DNA was discovered by a woman, Rosalind Franklin.

          4. Simone

            This is the most motivating (for me) piece of information ever. I thought of technology – both as work and field of study – attractive, because it is so intellectually stimulating. Thanks to your comments, now I also find it romantic, this idea that it needs both Y and X to be complete. I would much more prefer this sales pitch to women considering to pursue technology, as opposed to a feminist propaganda that I find off putting (come join because you are too few in tech; come join because you can do this too).

          5. Simone

            I can’t shake the beauty of your idea, I know very little about the topic, about to find out more. Your comment reminded me of a chat I had recently when the other person was referring to ‘femininity’ and I thought she was referring to motherhood etc, but she clarified she was referring to creativity. Seems that most of the innovation so far in AI succeeded in the calculus/analysis/statistics fields, as to mirror the creators (mainly guys).Perhaps you are right, perhaps this is the reason why they didn’t figure it out yet. Or maybe this is the foundation or one of the halves to make the whole.

          6. Twain Twain

            AI’s largely been made with male Y code because:(1.) It’s not a field where there are lots of women(*).(2.) Maths and physics are core tools and women tend not to study maths and physics — much less also code.(*) Interestingly, the women who are in the field all invented the emotion frameworks for AI (e.g.….

          7. Simone

            Thank you very much for the links provided, I have saved them and will review in detail for the writing project that I have joined. Can I ask just one more question, as this came up early in my project – is there a formal nomenclature covering the AI field? I find confusing in most media the use of terms like AI/machine learning/deep learning/robotics etc, so I could really use to know the academic nomenclature for all the related terms, if it exists.

        2. creative group

          Twain Twain:There are many Billionaires who actually don’t know Quantum Theory that is why they are paying people who do. Many the names you cited didn’t know sh**t about Bitcoin but their money gave them access to the head of the line in directing the mining of bitcoins.Read Digital Gold by Nathaniel Popper.

          1. Twain Twain

            It’s been the case for centuries that the world and its artefacts have been built by and in the mind moulds of the men with the money.Whether it was the Pharaohs and their pyramids, the Emperors and their Great Walls, the Carnegies and Rockefellers or today’s tech billionaires.With the exceptions of Queens Elizabeth I and Victoria and Baroness Thatcher, empire building and forging economic systems have largely been male because of access to capital and its asymmetry.Is it paradoxical that an elite handful of super-rich American libertarians are financing the creation of Blockchain and the mining of bitcoins to decentralize data globally and put the power of that data back into the hands of people around the world? Is it right? Is it wrong? Is it self-interested? Is it “for the greater good”?No one knows anything for sure.All of us are imagining and experimenting through life — whether we’re billionaires or little ants.I’ll put Popper on my list, thanks.

    2. Mark Essel

      Spot on, but I see AI or processing power being a new form of wealth. Open access, but not equal access.

      1. Twain Twain

        Pls see comment on the billionaire camps of AI. It’s definitely not equal access because 50% of the world’s population haven’t been involved in the design and engineering process — whether that’s Blockchain, AI or Ethics of Algorithms.And, well … unless some modern-day Ada Lovelaces solve the code for Natural Language …Today’s Babbages (Page, Zuckerberg, et al) can build bigger and faster processing machines BUT THEY STILL WON’T BE ABLE TO GET THE MACHINES TO UNDERSTAND OUR LANGUAGE, MEANINGS, VALUES & ETHICS!:*), LOL.

        1. Mark Essel

          Language is tricky. It takes time to learn to communicate efficitively with others (you learn their patterns, approaches, body language, style and vice versa). Machines require empathy to communicate

          1. Twain Twain

            A better language model for the machines isn’t so tricky once a person grasps the problems are to do with probability.@fredwilson:disqus — To achieve Ethics of Algorithms and true Machine Intelligence where the bots understand us, our language and our values to be able to allocate resources intelligently (unions, benefits et al) …Some BIG WRONGS in the toolsets we’ve inherited since 1654 will need to be righted and fixed.Yes, the smallest MVP invention can right 360+ years of wrongs and get us to Ethics in Algorithms and better Machine Intelligence.

    3. pointsnfigures

      thanks for the blog topic. Businesses that do well empower individuals to maximize their own individual utility. Businesses that can execute on that concept get much bigger than collectives.

      1. Twain Twain

        So true, thanks. Google became Google because its search bar maximized people’s ability to find things. The advertising model was an addendum rather than their original raison d’être which was basically just to help people find things on the web.Ditto FB: helping people connect with their friends.It’s a really subtle lever in economics is individual utility that compounds to bigger than the collective!

  6. JimHirshfield

    Why no ad tech? What happened there?

    1. jason wright

      is it native to the future web?

      1. JimHirshfield

        That doesn’t necessarily mean anything. Nonetheless, bait taken. It’s as native to the web as any other medium.

        1. jason wright

          advertising will survive on the web if advertisers directly pay consumers to consume it, in Satoshis. the Google model could fade out quite quickly.

          1. JimHirshfield

            If advertisers pay consumers to interact with their ads, that’s not advertising; that’s bribery or flushing money down the toilet.

          2. jason wright

            the fusion of the bit electron to the content electron.content downloads with a micro payment attached, a programmable micro payment, with variable terms and conditions. just the beginning of a whole new way.

          3. JimHirshfield

            I’ll line up my army of bots to participate right away. I’ll make billions, brahahaha…

          4. jason wright

            this is a peer to peer flat earth environment. this is not a pyramid.

    2. fredwilson

      Don’t like the economics and the commoditization (maybe the same thing)

      1. Matt Zagaja

        Neither do many of the users (well the advertisers seem to like it, the people supported by the ads are struggling). The weird thing is that you’d expect content producers like BuzzFeed or Verge to eventually just fold. YouTube creators would stop making YouTube videos because it doesn’t pay. Content seems completely impervious to the rules of capitalism and market forces and I don’t get it.

        1. Drew Meyers

          I’m not sure how long it will take, but I think we’re going to see a shift away from the ad model.

        2. ShanaC

          creating stuff has a market drive, it is more than ad space of quality is hard to price because of a mixture of commoditization of ad space/inventory types, unclear math behind branding -> lifecycle of buyer behavior, the sheer presence of luck for any piece of content to measure its value at any given moment, and FOMO of the “right audiences as leads to other audiences/buyers” by buy side.Much better math behind planning, much better tracking, and better ways of buying would actually kill off some media pretty quickly.

      2. JimHirshfield

        Well, I’d like to get your advice on a related topic. 😉 No, seriously, I would.

      3. JimHirshfield

        Well, I’d like to get your advice on a related topic. 😉 No, seriously, I would.

      4. ShanaC

        Yessss and noooooThe economics are really poor. Commodization is clearly happening. The slate needs to be cleaned out – mostly so we can figure out how media companies actual work (again) especially without VC backing (again)Then we can get into what kind of networks they do and don’t have

      5. Lawrence Brass

        Neither do I! It is an issue and I guess it would be great topic to discuss (or revisit) here at AVC as Ad tech and advertising are so relevant to build revenue and, on the other side, so detrimental to UX. Even banners looked elegant compared with the mess we have today. What is USV advice to portfolio companies on this matter? Can that be disclosed?Thesis 2 looks good, there is so much to do just below the application level, we have been gluing things together almost carelessly for too long. IMHO of course.The site looks good too, the only minor issue I had on an iPad ocurred when trying to expand the articles. Instinctively touched the block first, then the ellipsis and at last the title, only the latter worked.

  7. pointsnfigures

    like the web redesign. clean, educational, and to the point.

    1. Nick Grossman

      hey Jeff — thanks, and thank you (along w many others who were regulars in the last iteration of for joining the conversation w us. We hope that this iteration will still be open and welcoming and interesting, and will also be something that we can all keep up with a little better

  8. William Mougayar

    Congratulations on this evolved articulation. But the outside links contributions are still enabled in this section:

    1. Nick Grossman

      Hey William — thanks, and thank you for all the contributions in the last iteration of the site. For this one, we wanted to leave it open as before, but give it a little more structure and set a more sustainable pace

      1. William Mougayar

        right sized contributions, not firehosed 🙂

        1. Nick Grossman

          Yes that is the hope!– on the fly

  9. kenberger

    you have a Nomad Stack!? wow, I am nominating myself mayor of that one, given my family’s ongoing work lifestyle!(PS: I wonder if some folks will think it to mean re advertising industry: North Madison Av) ;)Love the displaying of the stacks on, as they are MUCH clearer now, and more likely to “honeypot” relevant folks from around the world.

    1. jason wright

      but too focused on “Millennials”

    2. aweissman

      glad you noticed this easter egg Ken 🙂

    3. Matt Zagaja

      Nomad stack is an interesting idea. As someone who currently “lives” in two states I think the nomad stack could use some sort of service that makes things like taxes and vehicle registration optimization easier, etc. Of course some of the more crazy stuff to tackle includes the fact that health insurance is somewhat state specific.

      1. kenberger

        Yup. I’m an American living between multiple countries these days. Lots of interesting obstacles to get tackled in years to come, for the growing number of folks also doing this.

        1. Simone

          Another nomad here, my 3rd country and not the last one. I may even change continents.The kind of service Matt suggested may be a pivot idea for the global payments start-ups who have collected lost of local knowledge. My dinosaur bank keeps on advertising almost free international payments lately, so I am a bit concerned for this segment of fin tech, this is how I came up with the pivot idea.

    4. Drew Meyers

      How do you source housing opptys as you go?

      1. kenberger

        that’s among the greatest enablers/trends on the list that makes this sort of thing ever-more possible these days: airbnb, and improvements in search results from old incumbents like kayak, plus competition from hoteltonight etc.You really can pick great places for a couple night stay, the same morning. Or have your pick of a family place for a summer. Sure, there have been sites like VRBO, craigslist, and many others for a while, but it’s just the past few years that tools have brought such huge market liquidity and ease of use that makes them a viable part of day-to-day living.

        1. Drew Meyers

          I lived as a nomad from 2010 to 2015… and been working on a private hospitality exchange platform called Horizon. More couchsurfing territory than airbnb, but overlap with both long term.

          1. kenberger

            feel free to get in touch with me re Horizon, i’m always intrigued by innovators in such fields.

  10. William Mougayar

    What I like about it is that it’s malleable enough that you can drive it around almost anywhere you like. But it might require a bit more pattern matching recognition that the previous one.

  11. Girish Mehta

    This redesign of should be some sort of benchmark for a “topic-centric” VC site meant to showcase thinking.Aligns very well with self-identifying as “We are a thesis driven firm”.

    1. fredwilson


    2. Jonathan Libov

      Thanks @girishmehta:disqus. We’ve gone through a few revisions recently and this one is the first that really feels right.

  12. RichardF

    interesting to see there is no explicit (or obvious) follow us on twitter button and that your video and talks section is buried below the fold with no link to that type of content in the menu.I think a16z are crushing it with their content, particularly the podcasts.

    1. Nick Grossman

      that’s a good point — we need to refocus on out twitter presence, which we have frankly ignored for the past few years. the videos and talks is a new thing which we’ll be adding to. def agree that the a16z content is super good

    2. Simone

      Agree about a16z and would add First round capital for articles, I have a hard time to resist reading them as soon as they land in my inbox

  13. guest1

    this community has the same people commenting the same type of things every day. Where has the entertainment gone? I say all the regularers should leave or at least stop commenting. I think it’s time to say Bye to AVC since it very much looks like the blog is saturated and congested. Sorry Fred, the community is your best friend, but right now your worst enemy.

    1. ShanaC

      how would you change behavior to bring in new people

      1. Simone

        Hi Shana,Perhaps some healthy self-regulation:)-The ‘regulars’ need to treat this blog more like a public arena and less like their FB page. To elaborate, there should be less ‘between friends’ engaging in the favour of engaging the occasional visitors.- Often and easily people get distracted from the post topic and go on rather long debates on what they ‘really’ want to talk about (like this was their FB)- To a certain extent, I understand many people have a business agenda to pursue coming to this community, but some are here purely for the intellectual pleasure (hopefully most). I would like to see less comments wooing Fred; by contrast, I appreciate long posts from some regulars that add lots of value and I got to learn a lot- Also, I could do with only one father figure for this blog, the biological father 🙂 – Fred.But there is nothing you can do about it, it can only be self regulated.For balance, I admit that any new/sporadic commentator, needs to put in effort to become part of this community.

        1. ShanaC

          It is a big community, to some degree on a personal level I don’t want to stop the “FB effect” though I understand why others would (my pet theory is ties people here)I guess the larger question is, what can we do as a group, to get people to stick around when they post for the first time, as we;; as be friendlier to people on the edge (hi edge people! comment, I’ll talk to you!!!)

          1. Simone

            The FB effect comes with the price of edge people feeling excluded, so what would be a healthy balance of interaction?I have 2 suggestions. 1. I have noticed that certain topics attracted a much larger and varied audience/commentators. So perhaps an idea would be to lure this silent audience regularly, e.g. once a month via more general topics, where people who generally don’t comment feel more comfortable to speak. Perhaps many people just read but don’t comment because they think they should have something really smart to say. Nonsense, look at me :)2. I receive in my inbox newsletters from a16z and Firstroundcapital, so they feed me all their content in one go. While I am aware that USV produces much more content, I still only check this blog daily because this is how I started. An email bundling all USV generated content and presence on twitter should help refresh the readership here too. Did you ever come across those exercises at work/school when groups of friends are forced to associate only outside their comfort group, that was happening for a reason 🙂

          2. ShanaC

            All the time. I will pass this on more directly to Fred

    2. Simone

      hi guest1, please see below. did I cover everything? do you have other suggestions?

  14. LE

    So this summer Jonathan and Nick started hacking on another idea, which is to focus the home page on the topics that are most interesting to us right now.Nick and Jonathan the new site looks great (both mobile and desktop). Suggest as “good housekeeping” you file US Federal Trademarks for USV as well as the USV logo.Edit: Not to mention “AVC” and Joanne should do “Gotham Gal”.

  15. Saul_Lieberman

    Why show ony active investments?

  16. Guesty McGuesterson

    I’d like to be able to follow the USV homepage in a feed reader but the feed at looks like a feed of submitted links, several per day.Since you’ve organized the site as a set of main “currently top of mind” topics, a feed connected to that would be cool. Maybe it could update when a new topic is added or one of the existing topics has a significant update. One update every day or five might be a manageable volume.

  17. Joe Cardillo

    I really like the new site & thesis. On a more general level, I thought it was interesting (and telling) that you consciously + continously develop and then publish the thesis. It’s really easy to keep things in your head, and when you don’t get them out and published (whether more seriously or simply on a blog you don’t expect anyone to read), you’re cheating yourself out of the opportunity to develop how you approach. The simple, directly stated mission or vision of an organization is really important, either because of what it says or fails to say.

  18. sigmaalgebra

    When I hear about an investment thesis, I am skeptical. Still I could see some value in the USV thesis of a large network of engaged users in consumer Internet services differentiated by user experience and defended by network effects — some value but likely not a lot beyond what the particular investments offered without thinking about a pattern or thesis. Now that I just read Andy’s latest description of a thesis, I am still more skeptical that a thesis can add anything to just what the entrepreneurs bring USV.Today Steve Blank sent e-mail on his ideas on looking at the future by asking a lot of highly accomplished people: His point was that those people would mostly just talk about the past. So, following their advice would be just repeating the past, and that is not a good idea in an area of rapid change were disruption is likely a big point.For a thesis, I can see “the Internet”. Beyond that, no: Saying more in a thesis is too darned difficult. Or it is tough enough to think of another Google, Facebook, or Twitter; doing so with a thesis from 50,000 feet up is way too hard. Instead, just have to leave the patterns to the technology historians. Or, beyond just computing and the Internet, what the heck was the one pattern where Microsoft, Intel, AMD, Apple, Cisco, Juniper, Yahoo, Google, Facebook, Twitter, Instagram all fit? Right, there isn’t one.Or, for a military analogy, by 1940, battleships still looked like something big, mean, powerful, important, etc. Ask nearly any of navy experts then, and they would explain how great, central, essential battleships were.So, Germany build the Bismarck. Impressive, right? Could fire a shell of 1000 pounds or so 20 miles or so, and one good shot sank the HMS Hood. Fearsom, right? Not really. Instead, the Bismarck was defeated by some flimsy biplanes of wood, fabric, and dope.Why? Because of range: The biplanes had range of maybe 200 miles instead of 20. The biplanes had intelligence, a pilot, and could see, steer, and aim while the shells of the Bismarck were just ballistic. The biplanes carried torpedoes which basically were better weapons at sea than either bombs or shells.For more, at Pearl Harbor the Japanese were smart — they brought six carriers, and to heck with the battleships.At Midway, again, the Japanese were smart — brought four carriers with Yamamoto on a battleship in the back that never heard an explosion. And the US was smart, three carriers — to heck with the battleships.And the US victory at Midway was one of the most important naval victories in history because the US sank all four Japanese carriers which, essentially, broke the back of the Japanese navy and was, really, the turning point for Japan in the war.So, to heck with battleships. Instead, in 1940, pay attention to some flimsy biplanes with wood, fabric, and dope and build forward from there. Next steps? Sure, more range, fighter cover, anti-aircraft artillery, radar, including on the planes. Everything new; next to nothing from the past. Original. No pattern. No thesis. That’s disruption and an example of what Steve Blank was trying to say.People have been doing ocean navigation from the beginning of ships at sea. By the time of Newton, IIRC. there were, or were the beginnings of, clocks good enough to solve the longitude problem.Later was radio navigation and inertial navigation. But for the US missile firing submarines, something better was wanted. So, some physics guys at the JHU/APL got out an envelope and wrote out a solution on one side — satellite navigation. It was built, and the receiver on the roof of the building routinely found its position within a foot (for a while I was writing software for fast Fourier transform analysis of passive sonar data but in the group that did the physics and was doing the computing for the orbit determination and heard the stories and met some of the people). Yes, the later USAF GPS was better, within an inch or so, in 3D. So, no thesis or patterns. Instead, something original and disruptive. And mostly just ignore the navigation expertise from the past.Or, when Oppenheimer wanted to build a bomb, he paid nearly no attention to all the bomb experts with decades of experience. Instead, Oppenheimer, who likely had never built or even seen a bomb before, did something very new and disruptive. No pattern. No thesis.It’s tough enough to find a good, powerful, valuable, successful disruption: Finding a pattern or thesis that can predict several cases of some such is just too darned difficult. Instead, look at the projects the way people have since at least Newton — something new, one at a time, on its own merits with no thesis or pattern and little or no reference to the past.

    1. Lawrence Brass

      I think that experienced and successful investors, like Fred and his partners at USV, develop a sixth sense that people on other lines of work or on the other side of the table just do not have. Imagine the amount of information they gather every time they meet with investors, entrepreneurs, portfolio partners. They capture trends. Even bad investments leave behind a pattern of failure, that is information too. In my opinion, the thesis is a distilled writeup, a conceptualization of all that information to remember for future deals, and not the other way around. Perhaps ‘thesis’ is not the exact word to describe what it is.

      1. sigmaalgebra

        > They capture trends.So, craft beer is a trend. So, rush to get into the craft beer business, as bath tub brewer number 51? Gee, private equity guys, may be some roll up opportunities there?So, generally would want to be the first in a trend, e.g., Xerox with electrostatic photocopying, Intel in microprocessors, Microsoft in end user GUI computing for document processing, Cisco in IP routers, Juiper in BGP routers, Apple in smartphones.But being the first in a trend is not always good: E.g., if social media computing is a trend, then Facebook was not nearly the first and, instead, the trend goes way back at least to BBS computing via dialup and 300 baud modems, before that, before computing, to open air markets, back to gossip, back 40,000 years, and recently back through Yahoo Groups, MySpace, and, finally, for whatever reasons, Facebook. Then the trend has continued with, Twitter, Instagram, PInterest, SnapChat, maybe more, maybe all based on the axiom that there is little that interests people more than other people?So, the social media business is easy money? Nope: Google’s first efforts flopped. Indeed, Google’s main success — that is keyword/phrase Internet text document search, e.g., digital version of an old library subject index card catalog with results sorted by popularity of referencing — is not really social media. Gee, maybe Page and Brin just don’t get social?Hmm …, trying to make a billion based on manipulating the emotions of teenage girls? Since I could never understand teenage girls, sounds like “riding the whirlwind” to me.

        1. Lawrence Brass

          I meant using trend in a mathematical sense. With the lively information and data they have I imagine that they can sense a point of inflection ocurring in some particular market before other people. But really, as you wrote earlier, it is hard to believe that there is a magical formula. The beauty of venture capital is that failure is compensated by success, at least in a healthy market, which makes risk taking viable. It is very darwinian if you think about it, for all the successful technologies and companies you mention, how many companies had to die? And curiously enough, companies may die but teams usually survive, and the dying company DNA is transferred to the next by these ‘failing’ teams, or transferred deliberately as in the Fairchild-Intel story.Social internet is different, I am also puzzled at times when trying to figure out how it really works. Indeed, it is basically satisfiying the natural need of humans to communicate, to share, I agree with you on this, this is not a trend. But it is more than that, much more. I think humanity is recovering its tribal qualities by means of this ‘enhanced communication’, we are gaining un curated awareness of each other individually and as a whole, we are gaining transparency in the sense that wrong doers get exposed sooner than before and good ideas are more visible. We had these, when we lived at a human scale in tribes or small villages and we knew each other. I wonder if it will always be financed through advertising.

          1. sigmaalgebra

            You are right.I don’t see how to see how “hot or not” had any reasonable chance to become Facebook worth $304 billion, and that is one heck of a huge gap in understanding. Similarly for the earliest of Twitter. Apparently some others had a tough time seeing how to see how the beginnings of Twitter had a good chance to become the present success. Maybe even Fred was unsure!While I’d like to have a way to know ‘hot or not’ with high accuracy for all seed stage or earlier information technology startups, my approach is easier: Borrow from traditional applied math, science, and engineering as exemplified by various projects with lots of originality, new value, etc. from both the commercial and DoD worlds. I.e., those projects can be reviewed, quite accurately, just on paper, when they are only on paper, that is, well before the VC seed stage. A way to say a lot that is accurate and relevant for a project just on paper? Sounds good to me.Gee, guys, outside of Silicon Valley, in engineering and national security, people propose essentially just on paper. That’s no doubt how GPS, and many more astouding projects, were proposed, reviewed, and approved.Silicon Valley? It wants a foil deck, with 10 foils with only 10 words per foil and then a demo — all the technical work done — or data on traction.Really VC is much like commercial banking and private equity, that is, invests in businesses based on some simple metrics — earnings, revenue, or at least traction. VC is not much like traditional applied math, science, or engineering investing in projects based on careful presentations just on paper that can be reviewed and approved with good accuracy.E.g., for the SR-71, Kelly Johnson gave his presentation, submitted his engineering, got the work reviewed and approved, and built the thing. In the Sand Hill Road world, they might have said: “You build and demo one, and we will fund the jet fuel for the first flight.”Net, DoD can approve with good accuracy just from paper, and Sand Hill Road can’t.Right, again, I’m talking only the crucial, core technology part and the DoD projects don’t have a market part. But the DoD does have something severe and like the VC product-market fit part — e.g., the SR-71 had to be and was product that did very well fit the real need.But, yes, the SR-71 was not trying to grab the emotions and eyeballs of teenage girls — no one is foolish enough to look for something accurate for that!I don’t want the horribly low batting average of the VC world; instead, I wrote out the core technology for my startup on paper, carefully, well before the first line of software.And my software does something nearly unheard of in the VC world — the new technology is really just some new applied math, not some new hardware gadget or routine software with a new user experience. Likely the closest Sand Hill Road comes to applied math is some uses of encryption.That some math might be original and new technology blows the minds, or glazes over the eyes, on Sand Hill Road. But, we’re talking information technology, right? And how even to define information without math? Then how to do powerful manipulations of data for more information without math? A biggie problem is, Sand Hill Road doesn’t have one of their patterns for new applied math.Semi-great: Looking for things that are new but via fitting simplistic, empirical patterns in the past. Gotta tell you, the world that is good at working with what is new, that is, research, doesn’t do reviews by trying to fit into simplistic, empirical patterns from the past. The review processes are severe but are based on much stronger approaches that fitting simplistic, empirical patterns from the past.Sand Hill Road doesn’t do such reviews. Neither does private equity nor commercial banking.

    2. Mark Essel

      The technology disruptions you covered (thanks, enjoyed the read) are bigger than unicorn businesses. They are breakthroughs or significant advances for mankind.Yet there are patterns even there, many scientists discovering breakthroughs simultaneously. I like to think of it as a water rising over a glass brim – it’s going to leak out somewhere once molecular cohesion can’t match gravity.For tiny businesses that become huge, I don’t know enough to say they can be batched into rational groups or patterns, but serial successful investors would have a better take on that. Even if there aren’t any statistical patterns, we’ll still see shapes in clouds, and images in white noise. It’s the shadows of the doors of perception that we see.

      1. sigmaalgebra

        > They are breakthroughs or significant advances for mankind.When we can!> are bigger than unicorn businesses.Gee, as my brother said about obscene profits, “those are the best kind!”.

        1. Mark Essel


    3. Twain Twain

      For investors and founders alike, the value of theses is that they help tune the telescopes (prisms of perspectives).We both know that scattering a bunch of startups on some XY-matrix plane and then doing some type of linear regression or probabilistic correlation to “pattern recognize” their similarities and differences are somewhat of a blunt instrument.Notwithstanding this … I’m pro-thesis. It’s a cornerstone framework from mathematical philosophy.Have thesis / random ideas about X => structure thoughts => set up systems to do experiments => obtain proof points => adjust experiment as necessary => validate thesis.

      1. Simone

        I like both comments. So I guess what Twain Twain wants to say is don’t take the word ‘thesis’ so literally

      2. sigmaalgebra

        Twain, to me you are correct — for a description of much of pure/applied science and/or engineering. But I fail to see your description as that of a VC thesis — in part because of the usual meaning of thesis and partly because of the examples of theses from VC from Union Square to Sand Hill Road.I omitted how the heck can have confidence that something original is solid or at all promising: Well, for, say, GPS, at’s a lot’s physics, supposedly with enough from general relativity that without it the errors in GPS would not be an inch or so but about six miles. So, the physics was crucial.Is such physics, math, science, engineering enough for a project to be promising as a startup? Nearly never. Instead for a thesis, might as well totally wear out that word, beat like a “rented mule” (@JLM), want (A) a huge market, e.g., 2+ billion Google users, (B) a must have product/service for that market, e.g., lust after Taylor Swift, (C) some effective barriers to entry. The math, physics, engineering, etc. can contribute to (C) and might be the crucial, core, secret sauce for (B). Also need to avoid a long list of negatives — CEO, nasty, abusive, on the sauce, “low energy”, free spender, no discipline, product/service needs a $3 billion (1940 dollars) Manhattan Project, etc.Likely some reasons for startup projects looking not much like GPS, shift register sequence radar signal encryption, or laser gyroscopes is that so far there has been a lot of money to be made with fairly routine software technology building on the Internet and now standard computing, the fraction of people able to do GPS like things is tiny, and could make a big bundle in the VC business with a background in just news reporting or selling for Intel.I did see what looks like a conceptual flaw: (A) Smartphone home screen real estate is in short supply; tough to compete with Facebook and Twitter for the eyeball time of users so (B) go for the infrastructure. Like saying no more gold left in those thar hills so we will retreat to the picks and shovels business? So, with no gold left, who are to be the people with money buying the picks and shovels?Ah, to heck with gold, just over the hill is the mother lode of diamonds or platinum! Or find something more attractive to users than Facebook? I mean, is Facebook really the end of what can be supplied to humans with all the rest only for robots? I don’t think so!

        1. Twain Twain

          Pure/applied science and/or engineering involves objects and atoms that don’t have any subjectivity and don’t engage in dialogues or interactions with the observer who set up the thesis and the experiments in controlled environments. Spectrometers and conical flasks don’t speak back.By contrast, the thesis of investors (as observers) is affected by diverse factors, elements and cultural ecosystems that don’t fit into neat, logical behaviors.Plus people (founders) speak back. They have subjective differences.Investors and founders have their own prior frames of reference which need not be quantitative, e.g. 90% of startups fail, 1 in 1 million startups IPO, 12X ROI, 20% user growth month-on-month, 3 times better user retention than ABC.No investor thesis has been able to include personal chemistry or the gut instincts of “There’s something in this somewhere. Can’t pin a figure on it but it FEELS like this could blow up into something BIG AND IMPORTANT.”Ditto no founder thesis has modeled, “Wow, the journey would be so amazing with these team members and those investors.”

          1. sigmaalgebra

            That’s right and why the solid math, science, engineering part is mostly just for the crucial, core, technological advantage and barrier to entry, enabling secret sauce.For the rest, need forces powerful enough to minimize the importance of the subtle, unknowable, delicate, inscrutable human whatever. So, need, to quote, a “big ass problem”, for a “big ass market”, with a “must have” solution, and there have, from technology or whatever, a solution that is definitely the first good or much better than anything else. Want no doubts or subtle stuff.The US DoD can go from first need to conceptual solution, breadboard solution, testable prototype solution, first production solution with high reliability and effectiveness — e.g., GPS, the Navy’s prior version, the atomic bomb, the hydrogen bomb, phased array passive sonar, infrared visibility, and much, much more, all with minimal problems with team psychology, having just the right flavors of tea in the little kitchen, etc.E.g., as we may yet see in the Mideast, if you are a pilot of a fighter plane, then there is definitely something you want to avoid at all costs — a US F-22. No way. Don’t want to be within 200 miles of one of those. Even if there are five of you — drop down near the ground, get behind some mountains, light afterburners, get away as fast as possible, and then land and run or just bail out and float down.Getting within 200 miles of an F-22 is just quick, sure, being burnt to a crisp, suicide. If you care about winning an air war, then an F-22 is a must have. Cost? No object: Spend $100 million for a lot of flaming junk or $1 billion for a few F-22s — the only choice.E.g., I plugged together my desktop computer with parts from Asus, AMD, Maxtor, Western Digital, Antec, NEC, Sony, Samsung, installed Windows XP Professional, the .NET Framework 4.0, IIS Web server platform, SQL Server, etc. all as must have, must work, rock solid, products, and, thus, get to concentrate on my work, etc.Rock solid solutions for really biggie problems.Biggie example: Document processing. That is, get rid of typewriters, graph paper, spreadsheet paper, etc. Then get output on daisy wheel printers, dot matrix printers, laser printers, color laser printers, color inkjet printers, PDF files, HTML Web pages, etc. For the must have part — consider going back to a typewriter, the non-electric kind, with an eraser or white-out. Not good, right?Secret: I never wanted to be a college professor, was for a while (as part of having a situation that let me take care of my wife in her long illness), never liked it, but, really, couldn’t do it. Why? As an applied mathematician, there was no reasonable way to get the word processing done. Right. Literally I was not able to be a professor because of the extreme difficulty of mathematical word whacking. No joke. The situation was beyond grim. In all cases, everything I wrote in math, the word processing was much more difficult than the math. No joke. MUCH more difficult. Grinding teeth more difficult. Really, literally cracking teeth more difficult — I cracked two, had a root canal and crown on one and later lost the other one. No joke.I wrote my undergraduate honors paper on group representation theory, and there were a lot of subscripts. I did the typing. To roll the carriage the half step, I worked my left arm so hard it was sore for over a year later. No joke. Eventually the arm healed. Mathematical word whacking is grim stuff.But now, mathematical word whacking is still a little more difficult than the math but is much better. But to have the better solution, I am using a lot of the full power of my computer along with D. Knuth’s TeX, some PDF software, and a recent, very, very nice B&W laser printer. Now I could turn out math papers at several a year and get tenure as a college prof, except I want to be successful in business, the money making kind.Lesson: This must have stuff is important and will overwhelm essentially everything else. With a must have solution to a big ass problem in a big ass market, nearly all the other subtle, psychological, intuitive, touchy feely stuff shrinks to insignificance, less important than the torque wrench reading on one nut of five on the left rear wheel.

        2. Twain Twain

          The beauty of the human mind is we’re infinitely ingenious and creative.That’s why Facebook and Google aren’t the natural conclusion of “what can be supplied to humans with all the rest only for robots.”

          1. sigmaalgebra

            Once again you shoot me between the eyes: I took decades of frustration, pain, agony, failure, wandering in the dark, bumping into the furniture, to accumulate data on human females, eventually found a light switch, begin to see, and worked up some explanatory theories that did well fitting my data, and once again you shoot down nearly all my conclusions.

          2. Twain Twain

            Nothing’s ever concluded. Everything’s still to be discovered and invented.

          3. sigmaalgebra

            That’s a bit pessimistic!Let’s see: The product of compact topological spaces is compact. That’s the now classic Tychonoff product theorem, and it is equivalent to the axiom of choice. Done. Amazing result.In a Hilbert space, every non-empty, closed, convex set has a unique element of minimum norm. Done. I do like this result — the “minimum” part can mean the most green dollars!Given the set R of real numbers, a positive integer n, and closed set C a subset of R^n, there exists function f: R^n –> R that is 0 on C, strictly positive otherwise, and infinitely differentiable. E.g. for C pick really bizarre sets, say, the Mandelbrot set or a Cantor set of positive measure. Done. I like this result if only because it settles a question posed but not answered in the famous paper in mathematical economics by Arrow, Hurwicz, and Uzawa. Otherwise the result is amazing but a bit far from making green dollars.Every complex, square matrix is the product of unitary matrix and a positive, semi-definite Hermitian matrix. That’s the classic polar decomposition, the key to factor analysis, IQ tests, lots of approximation techniques, the singular value decomposition, most of multi-variate statistics, the main sufficient condition for convex functions, and much more and a good candidate for the crown jewel of linear algebra. Done. This result is a central pillar of any value now from big data.Yes, we want ideas that are “new, correct, and significant” and new can mean “unexpected”, but new alone is not nearly enough since also need the correct and significant parts.

          4. Twain Twain

            Au contraire, it’s super-optimistic!Just so you’re aware, I took apart the positive, semi-definite Hermitian matrix because it’s one of the bottlenecks for why Big Data can quantify but not necessarily qualify data points I’d like.

          5. sigmaalgebra

            Polar decomposition?Okay.Since apparently the standard HTML elements sup and sub don’t work at Disqus, here I borrow syntax from D. Knuth’s mathematical word processing software TeX.Theorem (polar decomposition): For positive integer n and n x n complex non-singular matrix A, there exist n x n Hermitian H and n x n unitary U so thatA = HUThat is, somewhat intuitively, for the set of complex numbers C, function A: C^n –> C^n and, there, the ‘work’ of A is just, first, to rotate and/or reflect keeping distances constant and, then, stretch or shrink along mutually perpendicular axes. So a circle gets rotated and/or reflected and then stretched or shrunk into an ellipse. So, that is all a square matrix, that is, a linear transformation on C^n, can do. Similarly for the real numbers R and R^n.Proof:Let A^* denote the transpose of the complex conjugate of A.Let K = AA^*. Then K is Hermitian, that is,K^* = (AA^*)^* = (A^*)^*(A^*)= AA^* = K.Then there exist unitary Q and diagonal D = [d_ij] with d_ii > 0 such thatK = Q^*DQLetH = Q^*D^(1/2)QThen H is Hermitian, i.e.,H^* = (Q^*D^(1/2)Q)^*= Q^*D^(1/2)Q) = HThenH^(-1) = (Q^*D^(1/2)Q)^(-1)= Q^*D^(-1/2)Qand(HH)^(-1) = H^(-2) = Q^*D^(-1)Q= K^(-1)LetU = H^(-1)AWe show that U is unitary, i.e., that U^*U = I.U^*U = (H^(-1)A)^*(H^(-1)A)= A^*(H^(-1))^* (H^(-1)A)= A^*(H^(-1))(H^(-1)A)= A^*(H^(-2))A= A^*(K^(-1))A= A^*(AA^*)^(-1)A= A^*((A^*)^(-1))(A^(-1))A= II= Iso that U is unitary.Then we have A = HU for Hermitian H and unitary U. Done.So, in this proof, all we did is do a little matrix algebra using the standard properties of Hermitian and unitary matrices. That is, we didn’t have to bring in anything but just routine matrix algebra. For such an important result, interesting.The result also holds when A is singular, and can construct a proof similar to the above but with some more details.

          6. Twain Twain

            Ha! This is music to me as you well know!So what happens with non-positve integers and what happens where n(positive) x n(negative) x n(neutral)?

          7. sigmaalgebra

            The case of n a positive integer is the only one of interest.

  19. Brandon Burns

    Nice concept for Seems like a more streamlined and easier to control version of what the last iteration was trying to do.I wish I had equally positive reviews for the visual design, though. I’d expect a venerated venture firm to visually represent itself a bit more… refined. As in most areas in life and business, appearance means a lot.Also, I’d love to feel a sense of the evolution you speak of in the “what we’re thinking about” section, like maybe simply ordering the categories in order of most recent update, total updates, number of investments in each area, most recent investment in each area, or a small nav that allows you to play with all of the above.#constructivecriticism

  20. ShanaC

    I’m not surprised by the new Thesis announcement.What do you mean by infrastructure, and do you see data as a type of infrastructure piece

  21. jason wright


  22. Mark Essel

    Maybe open that data up on a time lag like 10 years if it can negatively impact other investments (signaling?), kinda like secret docs

  23. aweissman

    SimScale is a a “market specific network”

  24. awaldstein

    Niche in another word?

  25. aweissman

    maybe but niche to me (us?) implies small and I dont think these are (or aspire) to be small that way

  26. awaldstein

    The term does carry a lot of negative baggage with it.

  27. Richard

    Doesn’t the thesis of “the innovators dilemma” parallel the USV investment thesis?

  28. Nick Grossman

    don’t think there’s a strong reason not to open source it — in the last round, I think we went too far to productize the open source version, beyond what we could really sustain. So this time it would be less of a real “product” and more of an open code base.

  29. Mark Essel

    And yet without niche the world would be Walmarts and Amazons all the way down. Yuck.

  30. awaldstein

    The number of small, single focused online retailers this holiday season just doing a great job is a sign of a change–in what both of us would think the right direction.