Video Of The Week: The USV Berlin Conversation
For the second year in a row, USV did an event in Berlin in November and invited entrepreneurs to attend a moderated discussion. Our friends at Tech Open help us produce these events.
This year the USV partners who were able to attend this event were Brad, Albert, and John. The conversation was moderated by Ciaran O’Leary, who is one of our favorite VC co-investors in Europe.
Here’s the video of that moderated discussion. It’s long, at almost one hour, so you might want to chromecast this one to your TV and watch “on background.”
Comments (Archived):
When Brad talks. Smart people listen.
And Albert. I’d be interested in knowing more about how he homeschools his kids.
“Entrepreneurs make the world a better place. We just give them money.” — Albert Wenger.Classic USV.
Thank you Fred. Safe travels home.
“Connecting atoms and bits” — Brad Burnham.THIS. The thing about bits is that it’s a different mathematical form from atoms.Bits are the language and units of the machines. Atoms are the language and units of humans.Bits are based on probability & statistics. Its applications span search & advertising, networks, economic modeling, Blockchain, AI (everything from visual recognition by the likes of Clarifai to Natural Language Processing by the likes of Google Now) and Quantum Computing. This is pretty much the basis of the incumbents’ systems.So Albert and Brad are right to say it would be hard to compete with the Big 4 here. They have dominant network effects to extend in adjacent spaces for bits-based data.NOT in atoms-based data, though, because the maths of the atoms is different.That leads to different forms of search & advertising, networks, Natural Language in AI, Blockchain etc.Harder still is integrating the bits and atoms such that both mathematic forms and their resultant systems are COHERENT.That means the machines would understand us and our meanings and values better. That’s more of a “Holy Grail” than mere decentralized distribution.It’s do-able and code-able.
It’s interesting to me that all of the partners at USV would be considered “old” by Silicon Valley standards. Also, not all of them have classic “operator” skills. Flies in the face of of current VC conventional wisdom.On Net Neut, the real end game was a hierarchy decided for all of us. The FCC. That’s against the USV thesis.
My observations were quite different.The most successful people I know are the ones who are the most nuanced and shout less. More semi-colons than exclamation points.It comes with wisdom and success that is not always connected to age.
I think we agree on my first observation. The conventional wisdom I am hearing now is that in order to understand VC, you have to be under 40. You have to be an engineer. You better be a former “operator” of a company. I think that is bullshit.
Picking winners is not a quantifiable skill by experience.
But some subject matter expertise can help distinguish claims that are unlikely to be true.
Having subject matter expertise gives you a seat of the pants feel but also a bias. Not having that expertise means you are going to (or should) consult multiple sources and hopefully come to the correct conclusion. This is all an art anyway obviously all depends on specifics.Say someone wants to do a company in the wine space. They pitch it to Arnold who is quite knowledgeable. He doesn’t think it will fly and passes. They then pitch it to another angel who knows nothing about wine. He asks his buddy who says it’s a fine idea (or maybe not) and then proceeds to ask a few other people before deciding on whether it will work or not and if he should invest. My point is that to say that subject matter expertise is close to 100% significant would be to say that subject matter experts are close to 100% correct when they size things up and we know intuitively that is not the case.
no way to get around bias. just recognize it.
Of course it depends a great deal on what the definition of a winner is. This is not pregnancy.
I think we disagree there. Experience helps. As Brad said in the interview, seeing this before in the 90s will help make them better investors today, if they learn from the past.
Experience on its own right means little.Understanding what happens means little.Applying learning from the past to what you do in the future is all that matters really.That’s the the definition of relevance and wisdom in my opinion.
Depends of the space. I can see the that being an engineer and or an operator would give you a competitive advantage the investment space of materials, Pharma, or biotech.See theranos
This happens at the peak of every cycle. Probably a good indicator of the peak.What helps is diversity. My old jaded, engineering, operator experience helps in certain areas. But it hurts if it is not combined with a healthy dose young exuberant, optimistic, I don’t know what I don’t know dreaming.
> But it hurts if it is not combined with a healthy dose young exuberant, optimistic, I don’t know what I don’t know dreaming.Try building an airplane that way. When it tries to take off, I’d very much like to, uh, sorry, not be a passenger, but watch, right, from a very safe distance a long way away.Instead, how about, with a very unhealthy dose.Look, there are (A) millions of people throwing darts at the dart board. There are also (B) a few people really good at darts also throwing. But there are so many more people in (A) than in (B) that it looks like the best way to hit the center of the board is your “young exuberant, optimistic, I don’t know what I don’t know dreaming” wild throws.Similarly for hole in one on some golf course: While the PGA pros make many more than their share, most of the hole in one shots are from amateurs who were lucky even to hit the green, much less the pin.E.g., Kelly Johnson actually, really, designed the SR-71 which very much, very well did just what the heck Kelly designed it to do, first time, no re-dos, while Zuck with hot or not mostly just got really lucky.Ah, gee, now, if Zuck had spent more time in the Harvard math department, e.g., started with their course Math 55 (Halmos, Rudin, Spivak) and gone on from there, maybe he would have found for hot or not a better way to find hot than just the optimization technique of explicit, exhaustive, sequential enumeration!Or, okay, Zuck, given a normed vector space V over the reals and a non-empty subset A of V, then the normed vector subspace Span(A) is closed in the usual topology of the norm? Instead of hot or not, how about proof or counterexample? Then what could that have to do with finding hot girls? Could do a senior honors thesis on that?Zuck — gee, rhymes with luck!Now, outside that situation, sure, bet that the next hole in one is from an amateur, not a pro. But from inside that situation and trying to make a hole in one, sure, definitely be a pro, not an amateur.Okay?
Very bad analogy. Facebook is not an airplane.
Right. But you missed the important part of the analogy: Can we actually from the beginning design successful projects or are we stuck with just throwing darts and getting lucky? Important question.Airplanes show that careful design from the beginning can work quite reliably. So, we should attempt and expect more of that in information technology.Sure, tough to beat luck except tough to count on it!
And I would research the Wright Brothers if I was really using your analogy.
The Wright Brothers did some careful engineering, e.g., designed likely the first, useful wind tunnel. That effort is much of how they were able to design a plane to have desired lift, drag, weight, and propeller thrust, at the beginning.But they did make a mistake: They didn’t understand Reynold’s number that says how to scale physical size. So, their thin wings and biplanes really were the right stuff for the tiny examples in their wind tunnel, but with correct Reynold’s number scaling a single, thick wing had a much better ratio of lift to drag — also was better structurally, could carry fuel, etc. It took until the 1920s or so for aviation finally to figure out that thick wings were better. Indeed, when in WWII England sank the Bismarck, they did it with biplanes instead of single wing planes.Also the Wright Brothers did have their eye on a market, to make money, and publicity.Those two bicycle shop owners from Dayton did well: Poor Langley fell into the Potomac River.When the Wright Brothers packed up and went off to Kitty Hawk, a long train ride, they pretty well knew that they would be successful and get their desired publicity, firm orders, sales, revenue, and profit. They were not just guessing or just throwing darts.Neither was Edison as he was well funded to replace oil and gas lighting. Problems with Edison: He was too slow to see the advantages of AC versus DC. And he was too slow just to do some basic materials research if only just from the periodic table. So, it was Swan in England who figured out tungsten and, also, exploited a recent, quite good, vacuum pump from Germany. That pair is what saved Edison’s efforts at electric lighting. Fortunately, Edison did see that, in comparison with Swan, a much thinner filament, higher voltage, and lower current were the way to go. IIRC, Swan was mostly an academic chemist — in this case with some crucial insight. Edison was a well funded entrepreneur. The funders were Rockefeller, Ford, Morgan, or some such? Edison was not just tinkering around. He was the seed for GE, Steinmetz, etc.And, like Ruth did that famous day, both the Wrights and Edison were calling their shots — designing their projects carefully enough in advance to remove nearly all risk.Ah, once I went to a research talk about Ga-Al-As hetero-junctions. Gee, sounds like fundamental, pure research, far out stuff, high risk, wastes money, right?Nope. At the time, the bandwidth of the entire US voice telephone network was less than 40 Gbps. But with a Ga-Al-As hetero-junction, could make a tiny, powerful, dirt cheap, solid state laser that could put 40 Gpbs on one wavelength on one fiber for hundreds of miles without regeneration, a fiber with some dozens of such wavelengths, in a cable with maybe 144 fibers. It’s the Internet backbone. The talk was by some guys at Bell Labs. They were not joking, were fully serious, intended wild success, and achieved it, as intended.Innovative, ground breaking, disruptive, successful, very profitable projects really can be designed carefully in advance and, then, executed successfully from just routine effort. Edison, the Wright Brothers, Kelly Johnson, Oppenheimer, Boeing, the designers of Keyhole (like Hubble but before and aimed at the earth instead of the stars), the JHU/APL for the Navy’s version of the later USAF GPS, and many others knew this — Silicon Valley is still learning.
The Wright brothers story is fascinating. I had the luck to visit the National Air and Space Museum in Washington DC recently and had a near religious experience staring at the 1903 Wright flyer, in an impluse I actually touch it. The model wings used in the wind tunnel were there too, very tiny indeed. You wrote that they did not understand the Reynold’s number, but to be fair, is there any evidence that they knew the Reynold’s number and the concept behind it by then? Wasn’t too early for it to be of general knowledge?I agree with you in what you say about careful design and actually do favor meticulous technical approaches to solve things, even when it takes much longer. However, I have never seen a project completed by means of ‘successful execution just out of routine effort’ as you mention, at least in my mundane experience. There is always some discrepancy moving from theory to reality.
I started my career around DC and liked going to the Mall, especially the Air and Space Museum. The rest — Hirschhorn’s, the National Gallery of Art, Mellon’s annex to it, the one on Technology, also great fun.> any evidence that they knew the Reynold’s number and the concept behind it by then?I’d have to look up Reynold’s number in the history of the Navier Stokes equations of fluid flow. I had a guy at Brown’s Division of Applied Math fervently, urgently, sincerely advise me: “Stay away from the Navier-Stokes equations”. Once at a Navy lab I did work with those equations a little, but mostly I took his advice so would have to look up Reynold’s number. However, apparently computational fluid dynamics has by now done better than some would have hoped.But, to be a tough guy, there was no law against the Wright Brothers deriving Reynold’s number on their own!Oh, yes, in the implementation, there can be problems, but for some of the accomplishments I mentioned those problems looked comparatively routine.E.g., IIRC Oppenheimer discovered that making the plutonium he needed was more tricky than first guess: There was something about the rate of the reaction that varies the proportions of two radioactive isotopes of that element. IIRC, the unwanted isotope could have put out enough neutrons to light the explosion too early, before full compression, and result in a comparative dud. J. Wheeler said how to tweak the reaction rate to get the proportions wanted. Supposedly the US Navy is a real stickler: To reduce radiation exposure to the crews, makes sure the US subs have just the right, very pure stuff.In my project, after I did my derivations, the rest looked routine. Well it has been. But there was a real unexpected chuckhole in the road: I wrote the code using Windows’ .NET, and there I ended up with 5000+ Web pages of .NET and SQL Server documentation just from Microsoft’s MSDN site plus some hundreds more pages from elsewhere on the Internet. That part took only “routine” effort but a LOT of effort and was unexpected and really set back my project. Past that now, though, and see no such thing ahead.What I am saying won’t fly on Sand Hill Road. There the investors are looking for patterns, even where the ones they are looking for are silly, and they just won’t look at the most powerful pattern of them all illustrated by applied math, physical science, and good engineering. That Sand Hill Road attitude was a total shock to me — how investors in technology could evaluate projects only as, say, ex-newspaper guys, English majors, history majors, etc. But they do. How? Wait for and then look for traction.Fine with me; I’m past that now, and the flip side of the Sand Hill Road attitude means that they won’t fund any serious competitor of mine. And if by some bold stroke of lightening from the sky, if they looked at more in technology, they would look only at computer science, not applied math; my view of computer science is that it is really short on basic methodology, has run out the potential of the way too simple approaches it has tried, is a dying field, and applied math, with much better methodology and a much more powerful base of work, will win out. Sand Hill Road believes that because the software will run on a computer, the right stuff for the core technology is computer science. Wrong. E.g., current machine learning is building heavily on a lot of work of L. Breiman, and he was one darned good applied mathematician. How does it go? “If everyone agrees, then someone is not thinking.”Apparently the biomedical investors do look at the core science. Good for them. Apparently the IT investors want to make a billion with another UI/UX and SnapChat. Yup, guys, bet your money on a still more highly polished, jewelry-like smartphone with a slightly improved UI/UX.
If you think that the value of your project, product or the advanced math behind the algorithms may be miscomprehended by the typical Silicon Valley investor, you may be right just because you are thinking so. If you have core science built into your product that is not evident, you will have to make it visible somehow. I guess Clarifai did just such thing when they published the app Fred posted about a some days ago. They were making visible and showing off the value of the underlying technology. And well, collecting data too.
Thanks, but long ago I finally gave up on having any IT investor in Silicon Valley pay any attention at all to anything technical.My first effort was: I did some research on how to monitor server farms and networks for anomalies. In important ways, that work remains the total cat’s meow for that important problem and apparently is still the first and only such work.Well, when I had the paper written and in very polished form, I wrote various firms in Silicon Valley, enclosed a PDF of the paper, and explained the business potential. I got back mostly just silence but in one case a confession that they don’t know how to handle research papers. Later the paper was accepted for publication in Information Sciences; I gave a talk on the paper at NASDAQ HQ in Trumbull, CT; wrote more IT investors in Silicon Valley explaining again, about the acceptance and talk, enclosing the final version of the paper and heard back next to nothing.One firm in Boston said that they might be interested if I could also solve the diagnosis problem: No, not at all easily; that problem remains too difficult due heavily to, so far, the need for massive quantities of historical data and far too little such data available. So, I confessed that, no, what I had was really nice, elegant, new, unique, powerful, valuable work but was real and that similar work in diagnosis was for now too difficult. I advised them that if anyone claimed to have a solution for diagnosis as good as mine for monitoring, then they, the Boston firm, should just laugh.So, I had to conclude, Silicon Valley IT investors just will not, not, take seriously, consider, evaluate, etc. research papers. They just flatly will not do it.The firms that have made money haven’t done it, and none of the firms will. So, right, even Thiel who made a big splash claiming to want more than just “140 characters” wouldn’t pay attention but did have a Stanford ugrad student call me; we talked for a while, and I wished him luck as he returned to school in the fall. I advised him to take a course from Luenberger, not from computer science. Maybe also Diaconis.Silicon Valley IT VCs can say a million times each, and a million similar but different ways each time, that, yes, they are different, leading-edge, want things that are original, powerful, new, advanced, disruptive, etc. That’s all just smoke to cover up what they really want: A good business deal based on a very promising startup with good traction in a huge market.In effect, they make a Markov assumption: The core technology and the future of the business are conditionally independent given just current traction. That’s not completely wrong, and in some cases close enough to being right, but also tends to throw out with the bathwater just the very best babies they do want.It’s simple: Whatever else, no matter what, they will not, really cannot, evaluate core research, and they won’t try. I have to suspect that should they consider research in a funding decision, they would get strung up by their toes by their limited partners.Or, it’s a very old story and goes back to the “The Little Red Hen” in Mother Goose, literally: No matter what, there was no interest at all until the hen had fresh, hot, fragrant loaves of bread on her counter with eager, hungry customers lined up ready to buy.So, really, right, show them the core technology by having them connect and evaluate the work. But then they will not be evaluating the core technology, which as far as they are concerned could be a monkey randomly pushing buttons or mice running across a keyboard, and, instead, will just be evaluating what the chances are that 1+ billion people would like to use it. They want to evaluate the product/service, not the technology.By analogy, for the Manhattan Project, they would have said: “You build one and test it, and then we will form a syndicate to chip in for the gasoline for the Enola Gay.”Okay, that’s their view.But as an entrepreneur, I wanted to use some core powerful, valuable, proprietary technology as a crucial advantage — pick a problem that 2+ billion people very much want solved, a problem where routine software can’t solve it, find a solution that needs some good, powerful, original work in applied math, write the software, go live, get publicity and lots of happy users, run ads, and make money.Once I had the math done and nicely written up, the project could have been evaluated, but no IT investor in Silicon Valley would even attempt an evaluation.The Silicon Valley people do at times make investments. That is, entrepreneurs apply, and sometimes get equity checks. But as in the video in the thread today, an equity check, even seed, A, B, and C equity checks, are not success. I fully agree.And at times the Silicon Valley (SV) IT VCs do make huge piles of money. On average their returns are poor, so poor no good entrepreneur should want to work with them.But, for my work, and me, I have to conclude that they want to get a seat on my plane after it has already left the ground, has some quite significant altitude, and is climbing quickly, that is, for me, way, way too long after the plane has started rolling.I want to be more like the guy that started the Canadian, Web based, romantic match making service Plenty of Fish — long one guy, two old Dell servers, $10 million a year in revenue just from ads just from Google, 100% owner, grew to 70 employees, and recently sold out for, IIRC, ballpark $500 million. Never took a single penny in equity funding.Some equity funding could have helped my project. E.g., at times I could have called some experts in .NET, ASP.NET, ADO.NET, SQL Server, Windows Server, etc., gotten some tutorials, sample code, etc. Could have helped a lot. Just some well written documentation could have helped a lot, but in spite of the famous, sweating Ballmer screaming “Developers!”, he didn’t go to the trouble to document his product at all well for “developers”. Apparently Ballmer didn’t know how to read technical material well enough to know that the people he had writing didn’t know how to write about technical material. So, what I had to go through with .NET was a self-inflicted, unanesthetized, upper molar root canal procedure — very, very wasteful. But, I’ve done it now. I worked through the 5000+ Web pages of .NET, etc. documentation and have 80,000 lines of typing, with 18,000 programming language statements, nicely documented, running, timed, tested, and apparently ready to go live. It works on some test data on my computer.I need to type less at AVC, stuff in some good initial data, that now I do have, do a few routine things, go live, get publicity, users, ads, and revenue, and grow. Heck, e.g., just two weeks ago I finally saw how to get comma separated value (CSV) data on a bank account transactions so that I can track my receivables, or just have a standard accounting program pull in the values and give me a report on who has paid, e.g., via direct deposit, and who has not — dirt simple, although none of the people at my bank knew how!Before I could close a seed round, I should have revenue enough not to be willing to accept an equity check.So far I am 100% owner and want to keep that situation. I don’t want a C-corp and do not want to report to a BoD. I’d be terrified to report to a BoD, e.g., like the people I got either silence or nonsense from in Silicon Valley. I’ve never been able to communicate at all effectively with those people, and I would likely have even less luck with them on my BoD; so far, mostly I got just silence from them, but on my BoD they would talk, and that would be much worse. Then they’d fire my ass, before I got any of my stock vested. I’d leave with $0 from all my work. They’d put a stuffed suit in as CEO, and he’d kill off the company ASAP. It’d be a total disaster in every sense. But, that’s just what the SV IT investors do: They know just what they want — another SnapChat, and as soon as they discovered my project was not that they’d be thrilled to see me and my project tossed in the Pacific about half way to Hawaii. And if they thought that there was anything in my project important but not routine, say, Python programming, then they’s get totally infuriated — really, really angry. Really arrogant, egotistical, psychopathic people very much do not like being told that there is something important that they don’t know, even if they were history majors at Williams College and got an LLB at Harvard Law.Or, apparently the way to start communications with an IT VC is to wait until they call the entrepreneur from the traction the VC sees. Fine. If they call, I will very likely have copies of e-mail I wrote them that did explain what my project was and that they ignored. Then I’d have to explain to them that the record is clear — they can’t understand my project and, thus, I can’t risk reporting to them on my BoD. Thanks for the call, and have a nice day, but I need to add up the ad network checks I got last week,install three new servers for the sharding in my architecture and a faster LAN switch, and meet with a network consultant on network performance, management, and security. Then I need to call CloudFlare about DDOS, security, and performance.By the time the Silicon Valley IT investors are interested in me, I will no longer be interested in them.I just don’t see any common ground between me and Silicon Valley IT investors.
Here are a few suggestions:(1.) Investors have to persevere through many meetings before they secure their funds too — in the video, Brad said it was 80 or so.(2.) SHOW & share how your system solves a problem — if people can’t get it or use in 1-2-3 steps then either:* pare the system down until they do; or* admit your starting theory is wrong, move on and do something better.(3.) Seek out a business coach who can go through your pitch and improve it.
Thanks.By now, I have a lot done, and the things to do now are just along the track obvious at the beginning. That track is fully routine. Say:(A) Load into my data base the good initial data I have. Do more severe testing of the site and fix anything necessary, important, or obvious — any bugs, tweak the UI, more initial data. Write a little software to do first cut processing of the log file of the Web server, i.e., something better than the little program I have now and the little editor macros I used during the software development. Also do something for processing the log files of my three back end servers and maybe something for monitoring SQL Server. Copy the production software from its development directories to a production directory and do the usual, first things for site security.(B) Do an alpha test with friends. Set up an e-mail address for giving help and getting feedback. Some VCs would like to “play with” the site at that point; contact them and send them a URL, with a URL with just a dotted decimal IP address and maybe for IP port, say, 81 instead of the usual Web IP port number 80 my ISP has, for my just consumer service, blocked. Fix anything that needs fixing.(C) See if my Web server platform software, Microsoft’s IIS, can do HTTPS instead of just HTTP. Get a business connection from my ISP so that I have a static IP address and IP port 80 open. Pick a domain name and register it. Do a new site logo PNG file with the domain name. Look into getting a site certificate from a certificate authority.(D) Do a beta test: Set up a simple blog for announcements, feedback, discussion, beginnings of publicity. Invite people, e.g. at Hacker News, various blogs. Fix anything that should be fixed.(E) Get a tax ID and a business checking account; register the business name as a trademark; submit whatever papers my local government wants, e.g., doing business as; setup an LLC; for tax planning, touch base with the IRS and/or a local CPA. Likely pick and learn to use some basic accounting software. Contact some ad networks and get set up to use their ads, collect statistics, and get paid.(F) Go live; get publicity, users, and ad revenue. Take advantage of the big collection of ideas I have on publicity.(G) Build a server, get a router, sign up for Microsoft’s BizSpark program and get enterprise copies of Windows Server and SQL Server; get a tutorial on using Windows Server; bring up the site on Windows Server. Do what is needed as the usage grows. With the log file data and ad statistics, do some good enough first things with ad targeting. If there is much need for more server capacity, then just simple arithmetic shows that there will be much more than plenty of funds for more capacity. Execute the growth. The software and server farm architecture are scalable just via simple means, e.g., just sharding, to a quite significant business; so as the business grows, add servers and a better router and, maybe, a server farm network switch. Take over a spare room as the server farm; upgrade the house electric power with 200 A in the main circuit breaker box and an additional, at, say, 100 A, in the server room; install a suitable air conditioner. Sign up with a content delivery network for performance and security, e.g., against DDOS attacks. Tweak the software to accept user IDs and passwords so that users can get notifications of new results and/or subscriptions to new results.Continue to tweak the site software.Then mostly just grow.That is, with the technical work now done, it’s all routine; the long list here shows that, even with quite a lot of detail, the tasks are all just routine. Many, millions, of others have done such things; I should be able to, also; and I should be able to get some good, free advice from some people who have set up many Web sites. There is more, but it is routine also.Just from some simple arithmetic, a server farm in a spare room and a first commercial ISP account kept busy can generate revenue enough to be quite serious about this project — e.g., be able to hire consultants, tax experts, lawyers, pick a business location, maybe in NYS, maybe not, lease commercial space, etc.Then grow to a serious business, e.g., all I can get of 2+ billion users. So, get some commercial space; do some hiring; go international, and keep owning 100% of the business — look, Ma, no BoD!Maybe drive down to the southern tip of Manhattan and have a chat with the GS i-bankers about what I might do.But, we were talking about equity funding: Whether any of this works or not, as far as I can tell, Sand Hill Road and IT equity investors and I will have nothing to talk about — they want to get on my airplane too late. Maybe they hope that my company has five co-founders, each poor, each with a pregnant wife, and is desperate for some equity funding mostly without reading the term sheet. Or, by the time they are interested in me, I won’t be interested in them. In particular, I really, really do not want to have a C corp and report to a BoD.Really, what I’ve seen of Sand Hill Road and the bios there leaves me very much not impressed; there is next to no IT equity investor out there I’d want to hire for anything at all significant. I sure as hell don’t want to report to a BoD of them. To me, beyond just their green money, they look like trouble.The good work I’ve done, e.g., just the elevator pitch version of the business idea, the core applied math technology, the software to date, more details on the business, e.g., getting data, being the unique source of results users will really like, all look good. That work is what is special, what has promise of making money. The rest looks routine, things that have been done millions of times by others; I should be able to do them, too.I’ve had a little tangential side project typing in some math notes into TeX, have 45 pages. I tweaked my standard TeX setup a little to give output easier to read on my screen. Tweaked some editor macros I have for handling TeX input.The typing raised a little issue I hadn’t noticed before: I guessed that something was true and tried to prove it. Got a proof for the first two cases, but the third case looked like a mess, maybe even not true. Glanced at some books — didn’t see the issue addressed.So, I looked for a counterexample — presto, bingo, boom, found a nice counterexample right away, less than two minutes! Yesterday I typed in the counterexample.Now for the paper, have just one theorem and proof to add and do know how to add it, might add some intuitive remarks, and then will be done with that typing.Just intuitively, I thought that all the content was true; it is; it is nice to see a polished presentation to support my intuition. It’s been a few days more work than I wanted it to be, but it’s a nice little paper to have, for my project and more.Alas, no IT investor on Sand Hill Road would be able to appreciate the power and value of that paper or even be able to read it; similarly for 99% of the people in computing and/or computer science, including the profs. Indeed, as I noticed early on in my project, there was a CMU CS grad student who gave a talk at Google and claimed that some of what my project is trying to do can’t be done. He had four reasons. He was wrong on all four; I’d already found solutions to all four. The paper is a polished version of part of the solution to one of the four problems. Once at the US NIST, I was in an auditorium of about 2000 researchers all struggling with that problem and believing, like the CMU CS student, that there was no solution. I wondered about that even back then. I was right to wonder; there is a nice solution.Got to get back to that typing. Nice little paper to have — would shock a lot of people! Most of those people would get really angry with me; although I have a solution to that classic, old problem they have struggled with off and one for years, they would get really pissed at me because I showed them something they didn’t know and had long believed was false. They’d get pissed.For something like that, out alone in the cold, essentially the only one in the world who knows, where some thousands of researchers and technical workers would be pissed at me for showing that something they had believed in for years was false, it’s just super nice to have math where can write proofs, and to know how to write proofs, and to have TeX to make the writing look really good! Cute little paper!Really, more generally, the whole project is like that: For the crucial, core technical internals, I’m the only one in the world who knows. With highest irony, the very best projects should often be like that, but it is just such projects that Sand Hill Road will ignore until it is too late for their equity check.Yes, there are not many examples, especially from Sand Hill Road. But, as we know, they try to swing for the fences but actually get such a hit only about once each decade. That’s a heck of a poor batting average. In a decade, lots of ambitious projects get done but just not many from Sand Hill Road. We have, right, SnapChat.Yes, a week ago, athttp://avc.com/2015/12/cont…it was good to see Ben Evans talk about 7. Search and Discovery While there is no evidence that anyone in Silicon Valley, IT, practical computing, or computer science has anything like a good idea how to do that, that is, write the crucial, core software, it was good to see recognition of some aspects of the need!Ah, I’ve been out there alone for nearly each time I did something that proved to be good:(1) In the eighth grade, my handwriting was awful and, thus, my arithmetic just awful, so my arithmetic teacher gave me a D and fervently, one on one, advised me never to take anymore math!(2) In plane geometry, I found what looked like a novel solution to a relatively challenging exercise, showed the teacher, and she said “You can’t do that”. Later I discovered that I’d reinvented the recognized advanced technique similitude. Sorry, teacher!(3) I was working on some impossible software project for some engineers, discovered that in a request for proposal (RFP) they wanted a way to pass white noise through a filter and, thus, generate sample paths of a second order stationary stochastic process with a specific power spectrum. Gee, now there is a beautiful summary athttps://en.wikipedia.org/wi…Yup, it’s a Hilbert space exercise! So, I gave up on the impossible software project and in four days wrote and ran some software for the stochastic process work. I didn’t tell anyone I was doing this. Then I called one of the engineers at the customer, and we played with my software. Presto, bingo, my software house got essentially sole source on the work in the RFP.(4) In some more military work, the US Navy wanted some analysis. I thought I saw a way. But since there was little confidence in my getting a solution, they set up a parallel project where neither of us knew about the other. For the other project, no idiot could have come up with a worse design. My project, software, went through perfectly, pleased the Navy, on time (with an absurdly short due date, two weeks), and was later sold to a leading US intel agency (could tell you which one but then would have to …).(5) I derived some math and wrote a paper. My bosses claimed that the paper was not publishable. They were wrong: The paper published right away, without revision in a first class journal, the first journal to which a submission was made.Then that happened a second time.It is as if, if a project looks good early on to the bosses or investors, then likely it is junk.Doing work as in my project or even just that little tangential paper, it’s possible to get some increased respect for Trump — he’s out there alone. For his business, it’s just private — don’t look for it on the NYSE. Among the Republican candidates, he’s alone — much of the GOPe would rather elect Hillary than Trump! For months he was alone and taking from the MSM and pundits the worst insults possible to put into words — they were all really badly wrong. On the other hand, now he consistently draws crowds that fill all the seats, 8000+, SRO, in nearly any covered space there is.Likely: No doubt the guy we would really want as POTUS nearly everyone would like after the fact but nearly no one would like before the fact. Same for Sand Hill Road — the projects they would really want, early on they will throw away.Then on to (A)-(G) above, hopefully ASAP. I should be able to have a really good Christmas in 2016.Blind faith is no good — instead, at least for the crucial core, have some powerful, original applied math with solid theorems and proofs. Good stuff to stand on.
If BoDs are not your thing, you may be better off trying to apply to the Research arm of one of the big techcos.
Sure, if take equity funding from an IT VC on Sand Hill Road, then there will have to be a C corp., a BoD with at least some of the members investors, a term sheet, more details on the investment, vesting, lots attention from lawyers, lots and lots of time, money, and effort overhead from the CEO for all the BoD and C corp. stuff, enough so that the CEO mostly can’t do the job he needs to do running the company. Bummer.A C corp. is granted by the government, e.g., in the US, often the state of Delaware, to turn a business into a legal person. For this grant, the government wants a BoD to be responsible for the proper, with respect to the public, operation of the business. So, the BoD is responsible for more than supervising the CEO, checking on the running of the business, and serving the investors and any other owners but also for the business being, call it, my description, a good citizen. Fine for UGE.But, in the US, if don’t take such equity funding, then don’t have to have a C corp. or a BoD. Or, the US is just awash, east to west, north to south, in businesses, some quite successful, that are not C corps. and don’t have BoDs. There are millions of successful sole proprietors. Indeed, in my rough arithmetic, nearly every house in a good neighborhood, nearly every boat over 40 feet long at a yacht club, nearly every winter ski lodge in the mountains, and a large fraction of parents paying full tuition at high end K-12 and Ivy League colleges get their money from owning, usually sole owner, of a good business. So, people working for a salary or professionals, even in law or medicine, rarely need to apply. E.g., the head of the family might own 10 fast food outlets,10 combination gas stations and convenience stores, 10 medical testing labs, 7 new car dealerships, the main wholesale plumbing supply business in a city of 4 million people, three million square feet of commercial rental property, be a partner in a successful IT venture capital firm, own a successful local cable TV and ISP business, etc.Net, very much need to own a successful business. And of such US businesses, nearly none of them ever even thought of getting equity funding from a venture firm. A lot of these people got started in their successful business because they were the children of parents already at least somewhat successful in such a business. I never heard this from my parents, was never taught this in K-12, college, or grad school, never heard of it when I was an MBA program prof, never read it in a book or magazine article, but had to discover it for myself — it’s the truth. That’s just the way it is for family financial security in the US.My goal is to own a successful business. What business? Well, I have a good background and, clearly, plenty of talent, in applied math and computing, and those two are relatively promising now. So, look at what might be done with those two; look at the Internet; notice many months ago what Ben Evans said last week, that and much more; plan a business that takes advantage of what I have in applied math and computing, design and write the software, …, go live, get publicity, users, and revenue, grow, and be financially successful. Simple enough.Just the same as millions of business guys all over the US do except I have some advantages. E.g., my father in law raised chickens 40,000 at a time, working at hard labor about 20 hours a day, made in today’s money $100,000 a month doing it. My work is indoors at a desk — much less exercise but otherwise much better. If I can get users enough to keep a $2000 server half busy 24 x 7, then $100,000 a month will be easy for me, too. E.g., look at the traffic Drudge Report is getting, get an estimate of his ad revenue, and add that up — way over $100,000 a month. For now, there’s a shockingly large amount of revenue possible on the Internet from just one busy $2000 server.Don’t need a C corp. or a BoD for that. E.g., a sole proprietorship is enough unless want some liability protection in which case be an LLC. Not really difficult.And, don’t necessarily need equity funding: It’s too easy at IT blogs and news sites on the Internet just to assume that somehow venture funding, Y Combinator, etc. are a necessary part of such success. They are not. E.g., plenty of VCs confess that their checks are not for everyone. And some VCs advise that they want to invest in companies that don’t need the cash. Then, of course there is some question why an entrepreneur would take an equity check: Well, in some cases there are some good reasons, but just running a successful business, like millions of US entrepreneurs do, is just not one of those reasons. I definitely should not need an equity check, a C corp., or a BoD.
So you are walking Jason Fried’s path, why not? I suggest you to prioritize doing the formal parts of (E), getting those done in sync is tricky. I have already done most of it myself, but have really been a PIA. Next time I will get some help.It seems that you will be having a lot of fun during 2016, I wish you the best.
Re: your point (G) – (paraphrased) “install a server farm in my house”. No, don’t do that, seriously. There’s very few computing apps you can’t build much more effectively on Amazon EC2. I’m sure you’re aware of it but you seem have have dismissed it. Take it from someone who has done both home-server-farm and EC2 (and the latter is now running with over 52million users) EC2 is the only way to go. It may irritate you to spend a week or so adjusting what you have to run in the cloud, but two things to consider;a) If you intend to scale to be large enough to be minting money, at some point – fairly early – on that line you are absolutely, totally fucked with a home-server setup. Migrating it at the point that it’s creaking under load will fuck you even more because you’ll have so many more valuable things to be doing with your time.b) If your plans somehow do not pan out and the world doesn’t beat a path to your doorstep, you won’t have a room full of servers and wasted all that time+money getting your rig set up (and as a useful side effect, you’ll be familiar with cloud computing which is without question the way of the future; well, actually the way of the present).Seriously, spend a week or two and migrate it onto Amazon (or Azure or whatever). If you have a codebase totally built on Microsoft products (which I wouldn’t especially recommend to a newcomer but whatever- that’s time already invested) you can still make it all work.To be considering having an in-house (literally!) server farm in nearly 2016, is dubious in the extreme. Without wishing to insult you, there are almost no software applications left where making that choice isn’t a crackpot idea.Best of luck.
Thanks.(1) No Biggie.In case my house server farm is no longer big enough, moving to a second server farm in some suitable commercial space should be fairly easy. Then likely I should talk to my ISP and, after warning users, one dark and lonely night shut down the first farm, tell my ISP to start sending the traffic to my IP address to my new farm (right, their router just needs the new MAC accress?), all in, what, less than 10 minutes, and then continue with the second farm.Then move the equipment from the first farm to the second farm and start making use of that equipment for more in capacity.(2) Windows.You mentioned Windows and my being new:I’m new to Web servers but not new at all to serious software.I prefer Windows to Linux because I do know a lot about Windows and know only generalities about Linux and because of the Microsoft product line and their track record in high end sites.(3) Trade Secrets.On using a cloud instead of my own server farm:For the cloud, I want to keep my back end server code a well protected trade secret. So, I’m terrified of trusting a cloud provider.(4) Farm Arithmetic.My arithmetic indicates that I can do big things with one small room with the server farm on at most just three wire rack shelf units, each 18 x 48 x 72″, $100 each at Sam’s Club.Then, if that room gets too small, I’ve got two more rooms also available.If all that starts to fill up, then I will just get some suitable commercial space, while still running the first server farm, set up a second, larger farm in the commercial space, one with some professional engineering and not just my DIY work, and then on a relatively unbusy night, bring up the second farm.Then I’ll move the equipment of the first farm to the second one and start to make use of it.My arithmetic indicates that I can make a lot of progress keeping just one server for less than $2000 on average half busy 24 x 7.How much progress? Well past being willing to accept a Series A.Say that the server is in a cheap-o mid-tower case (beige, hopefully instead of black!), with an Antec 650 W power supply, an Asus motherboard, an 8 core (HOT!) AMD processor at 4.0 GHz, lots of fans in the case, 32 GB of Kingston ECC main memory, several hard disks, and one solid state disk (SSD) for a special purpose where and SSD is just terrific, and that’s about it.With my software, with the Web server, the session state server I wrote, SQL Server, and two back end servers, all that software, communicating via just simple TCP/IP sockets, all on that one physical server, I should be able to send an average of 10 Web pages a second.Looking at the Internet upload data rate, my Web pages are dirt simple — each sends for at most 400,000 bits. So, the peak data rate, 20 pages a second, twice the average, is 8 Mbps, that is, trivial for essentially any Internet connection.The mobile strategy? HTTP/HTML. Each Web page is exactly 800 pixels wide, large fonts, high contrast, dirt simple layout and UI, formatted with just tables, can look okay on a screen as narrow as just 300 pixels, next to no usage of JavaScript, no pop-ups, pull-downs, roll-overs, video, Ajax, screen jumping around, etc. A user’s Web browser needs to be up to date as of, say, five years ago. For the front end, we’re talking just dirt simple.Drawing from the Mary Meeker KPCB reports, for ad revenue, my planning includes a charge per 1000 (CPM) of $2. Each Web page will have one banner ad at the standard size 720 x 90 (W x H) pixels and an average of about 4 ads at the standard size 300 x 250 (W x H) pixels. So, that’s an average of 5 ads per Web page sent.Then at an average of 10 Web pages a second, the monthly revenue would be2 * 5 * 10 * 3600 * 24 * 30 / 1000 = 259,200dollars. Pretty good for one server with less than $2000 in parts. Presto, bingo, at that point I will have no more interest in equity funding. E.g., a few months of that and I will have cash enough to match a common Series A. “Look, Ma, I still own 100%, and no BoD!”.Net, by the time my home farm gets very busy, I will have plenty of cash for plenty of good options.The software and server farm architecture are scalable via just simple sharding. On a wire shelf unit 18 x 48 x 72, can get about a dozen such mid-tower case servers. Call that a factor of 10 in capacity for monthly revenue of$2,592,000So that’s from just 6 square feet of floor space, and I can have three rooms of space in my first server farm.And with that factor of 10, the upload data rate to the Internet would be 80 Mbps, still modest.So, plenty of cash for very rapid organic growth.About that time, sign a lease on some commercial space. Get some professional site design, e.g., with uninterruptible power with a Diesel emergency generator.Also, in the same building, get some office space and hire some people, first an office manager and COO candidate; then a server farm operations guy and CIO candidate, etc.
I believe you to be making a poor decision in terms of best use of your time, money and the QoS you will be able to deliver to your users. However you seem assured in your choices, and I don’t disagree that you’ll have an interesting and educational experience going your chosen route.
I’m concerned about the security of my proprietary assets to be protected just as trade secrets — that is, don’t tell anyone.If someone gets, midnight software house or whatever, my proprietary core secret sauce technology, then I’d have one heck of a legal and practical problem putting that toothpaste back in that tube. Better be sure the toothpaste just stays in that tube to begin with.For the rest, I believe that we are both right but are using some different assumptions.For QoS, initially, I agree, but I expect QoS problems early on no matter what. E.g., in part to minimize harm to users from QoS problems, early on users won’t get UIDs or PWs and won’t sign on. And there is no use of cookies. So, essentially a user’s interaction with my site from day to day is, right, stateless. So, if my site is down on Monday and comes up on Tuesday with some changes, then no user lost their state or any of their subscriptions, options, work, or anything else.For my time and effort, early on, learning about a cloud, at Amazon, Microsoft, or just my local colo shop, dual 10 Gbps connections to the Internet, blazingly fast, 70 miles north of Wall Street, likely with fiber optics down the Hudson River to lower Manhattan, would take time and effort. I’m saving on that. E.g., no premature optimization — no brick outhouse with heated bidet and marble floor?But I do have some time and effort I have to invest in production usage and system management for Windows Server and high end versions of the Web site platform IIS and SQL Server. Doing this with my first server, not in the cloud, not even in a spare room, but just at my left knee, is the way to go. I’d be using Windows Server in the cloud, and that better darned well not be my first use of Windows Server.I’m a solo founder, a one guy shop, janitor, coffee maker, secretary, security staff, accountant, junior and senior programmer, software team leader, system management bridge manager, CMO, CFO, CIO, CTO, and CEO. So, for now I have to be going from walking to a bicycle to a used pickup and be a long way from a high end Boeing 757.After even just two months of revenue from just one busy server, I can easily have a lot of sympathy for all you are saying, once I can get around the trade secret problem.Once my site is at all successful, if I am running on Amazon, Microsoft, Google, Oracle, whatever, then Bezos, Nadella, Page-Brin-Schmidt, Ellison will notice me; some staffer will tell, say, Bezos, “You, know, he’s running from our cloud”, and eyebrows will go up, smiles will spread, and the staffer will continue, “We’ve been collecting stats on his performance, and it’s amazing how fast his software is. We’ve used his site, and it’s different, very, very different, works great, still is easy to use, even by, say, some 10 year old off the street and in an Internet cafe in Thailand, but we don’t have even a clue about the internals of his software. I’ve chatted with some CS profs at Stanford, and they don’t know, either. It doesn’t look like anything in AI, ML, or big data anyone has seen.” Then with some interest and satisfaction, some guy will mention the obvious, “You can’t patent software.” Some other guy will follow up with “So, let him sue us. How many lawyers does he have?”. I’d rather just imagine such a disaster than have to live it or even sleep while worrying about maybe having to live it.Really easy step: Just don’t tell anyone, and keep the darned core technology locked up as tightly as possible in my own server farm, where I’m the only one around until I have money enough for a solution, say, much better than, uh, let’s see, say, just to pick a site, NSA.A room with three shelf units with servers, a biggie A/C in the window, and a spare ready to install just in case, a good UPS, a good Diesel generator on a concrete pad just out back, and the associated revenue for a few months would let me rethink this whole subject.The Diesel generator? About all it takes is just money. Pick some highly recommended people, take their recommendation, turn them loose, write them a check, and DONE. Suddenly everything appears and looks good — the concrete slab, the generator, the weather housing, the aluminum electrical conduit, the automatic switchover box, the weekly testing procedure, the fuel gauge, the wire to the UPS, the electrical and fire safety precautions, etc.And, if a cloud is then the way to go, sure, presto, bingo, get a good consultant from the cloud, upload everything — have lots of containers, virtual machines, e.g., make sure both my code and IIS are linking with the right version of the .NET Framework, ability to move a running program from one VM to another, some more in fault tolerance and scaling, a tutorial on how to do the system management — then click on a button for a full switchover. Sure, plenty doable when and if then.
You list so many things you will need to personally organize and manage/maintain (and omit many more) and then hand-wave all the work away. The sole argument you have against using the cloud is a fear someone will somehow steal your secret sauce if you use the cloud (with no logic to explain why it is any less secure than a homemade server farm). Based on well over a decade of experience doing this stuff (initally running an IIS & SQL server rig on a T1 in my office in 2001) and since then much largrr systems (52M user system running on linux boxes on EC2 since 2011 with less than 20H downtime) I conclude I am wrestling with the proverbial pig in mud, so I will leave you to enjoy it. Best of luck regardless, you should be clear you are making technical/business choices based on what toys you personally want to play with; it’s a good thing you’re not reporting to a CTO because this is narcissim not sound engineering. 🙂
Thanks for your interest and efforts.First, for your You list so many things you will need to personally organize and manage/maintain (and omit many more) and then hand-wave all the work away. No, not right: I definitely should be able to bring up my site.For growth, if my computing starts to get busy, then I will have plenty of options for growth.Staff? I can go a long way getting details by working with vendor technical support and expert consultants. E.g., I don’t yet have to have in my company an expert SQL Server DBA.Sure, I’d like a few minutes on the phone with Mark Russinovich.For your The sole argument you have against using the cloud is a fear someone will somehow steal your secret sauce if you use the cloud (with no logic to explain why it is any less secure than a homemade server farm). No, not the “sole argument”. I also emphasized that I need experience with enterprise versions of Windows Server, IIS, and SQL Server and their performance monitoring and management before I try to use those in important production on a cloud site 2000 miles away.So, maybe you have an argument you have not stated: So maybe Bezos and/or Nadella have people who would walk me through my code and data upload and installation, configuration, and production on their cloud where I don’t need to know more about Windows, IIS, and SQL Server than I do now. Maybe. I doubt it, but maybe. So, the cloud staff handles VMs, load balancing, response time monitoring, automatic sharding, backup, recovery, DDOS, more in security, site restarts after software updates (“do a switch-over and hope we don’t get a burnout”), etc. I don’t have to worry about the version of the .NET Framework, SQL Server connection pools, etc. Maybe. They provide kids with a real tooth fairy, too, right?So, right, I haven’t investigated that, and I have yet to dig in my backyard looking for Captain Kidd’s Treasure, either.But the bigger issue really is security for my intellectual property where the only protection I can hope to have is trade secrets — just do NOT tell anyone so that the only way they can get my work is just to steal it. If my work gets stolen, then my options are poor — as I mentioned, can’t put that toothpaste back in the tube. If NSA can’t, then I sure can’t.In more detail, for your a fear someone will somehow steal your secret sauce darned right they could. I did mention NSA. The threat is real. no logic to explain why it is any less secure than a homemade server farm If you have some information about how a well run, vanilla, basic Windows Server, IIS, SQL Server Web site connected to the Internet is necessarily at risk, then by all means be a White Hat and tell Nadella, Russinovich, etc.I have a heck of a strong background in computing: You seem to claim that I do not have enough background to be a responsible Windows Server user. That claim won’t wash.Sure, there are various risks, but IIRC Windows Server gets used in some quite serious sites.About your position, I’d like a few minutes on the phone with Mark Russinovich.Sure, Flash has long been a wide open vulnerability. Sure, send someone e-mail that they receive with Outlook, with a Word attachment, with some OLE data embedded, is a threat for a user dumb enough to open such an attachment with Word. Maybe some aspects of client side JavaScript are dangerous. No doubt a lot of applications software is risky — maybe some accounting program that automatically downloads bank transactions is wide open to, say, a MIM attack.But my Web server won’t be doing any such things.Yes, I’d like more in computer security, e.g., some actual guarantees of ability to run any software, especially malicious software, safely.Gee, first rule of computer security — never let data from an untrusted source be executed as software. Following just that rule would stop a lot of attacks. E.g., no way would I give a file from the Internet to Excel or Word — not a chance. Never have; never will. There will no copy of Flash on my Windows Server.I really don’t think that some hacker in Belarus will be able to get into and run their code on my Windows Server computer.Really, the usual main threat to computer security is from the inside, e.g., Snowden at NSA. Gotta tell you, I know with 100% certainty that there is no Snowden like threat to my company now.For a cloud, sitting here, I can’t be sure there is no Snowden-like threat. But at a cloud shop, not even a Snowden could steal my code on their cloud if my code is not on their cloud.Sure, once my company starts to get revenue, I will have some good conversations with CloudFlare, etc. And, there is at least one box can put on the main network link that will stop a lot of stuff, say, low level TCP/IP attacks.From banks and more, I tend to believe that basic server farm Internet security is now a well understood art. I don’t have to reinvent that wheel; a thick checkbook and some good experts should be plenty.For your you should be clear you are making technical/business choices based on what toys you personally want to play with; Not at all. I’m trying to minimize time with toys. Windows Server and SQL Server, those are not the toys I’m looking for.Microsoft’s products are not toys to me. Tools? Yes. Toys? No.You believe I’m doing bad “engineering”. No, not making any serious engineering mistakes yet.Sure a computer in 2001 might have had a single core with a 90 MHz clock. Now we’re talking 8 cores with a 4.0 GHz clock. So, let’s take the ratio:4000 * 8 / 90 = 356Hmm, so, first cut, a 2001 server farm with 356 mid-tower cases shrinks to one. And of course Intel has some 18 core Xeon processors, at 2.5 GHz, and can put two on a mobo for a ratio of2 * 18 * 2500 / 90 = 1,000So, since 2001, server farms can be smaller.Gee, there is a broad issue: Control. Like a really good car — one that runs well and is already paid for! Makes me sleep better.Put 100 people on this right away? Equity funding? A BoD? I’d be terrified. E.g., you already believe I should be fired.Really, given human nature, a lot of such negative feedback is inevitable. I will get criticism at least until I have a 200 foot yacht in Long Island Sound.Thanks for your interest and efforts.
not all of them have classic “operator” skillsImportant to mention that some or all of those “classic” operator skills are honed in one particular area or at one or two companies. That can be a benefit but it can also be a drawback. At least if you have no classic operator skills you have less of a bias and in theory you might be more likely to consider other points of view instead of relying on your gut (bias).
Wrong. You can choose to believe that but its wrong
The FCC is not a hierarchy?
Best hour I’ve spent in a long time.The most telling piece is the honestly during the pitch process. The understanding that no one has all the answers.We all need to hear and and listen to this hard.And understand that not all funders in my experience would agree.
The hidden investment thesis of only needing to be partially correct to the really successful within a single investment tell you how USV uses the goldilox approach of sector lifecycyles so well.My guess is that they look to see how the investment will sharpen that skill set as part of the decision to invest
This is a huge dilemma and is one many founders in my experience face. There are funders (I assume that wasn’t a typo) that really do want you have all of the answers.I think that comes from inner self confidence on both sides. If you both have it you can say: Good question. I really don’t know. Let’s talk about that. One way of looking at it is this, the other is that.
Not a typo and I agree.I recently assisted with raise that came almost entirely from family funds most that were a generation removed from the wealth creator.They wanted everything buttoned up. They wanted all the answers and of course, no one has them at an early stage.I’m with you on this and find it best to simply keep searching till you find the right people to go with the capital.
But here is what is hard: Sometimes perfection is the enemy of done. I hear people wax poetically about taking money from the perfect partner, and Fred might be that, but he is only going to do a couple of investments a year, and many he picks like Quizlet.So have I done it??? Yup. Am I proud of it??? Well, I was young and needed the money, don’t judge me. Did it work out??? Yup. Was it everything I ever wanted??? No, but I did well.
I’ve never taken any money from Fred.I’ve taken it from good partners and not so good,I’ve always been glad to have it.
Spoken with somebody from experience.
It depends on the stage of investment. In the early stages (which is where USV starts primarily), not all answers are known at the detail levels, which is why the quality of the conversations with the entrepreneurs is key, and a more telling sign.
Not exactly sure of your comment? We are in violent agreement. As a matter of fact I don’t think you know all of the answers in a late stage investment. You might think you do, but what if you changed this or somebody came in and did that/??
Right. The nature of the questions change over the lifecycle of the company evolution. We agree. I was merely reiterating, perhaps paraphrasing. What’s unknown becomes known, but new unknowns arise.
Leadership and clarity of conversation is no less important at $100M than at $10m William.I would state that actually it’s get more important as you grow as the cost of growth increases as you increase in size and as revenue determines your thinking more and more.Don’t see the distinction you are making here I guess.
Except that in the early stages, the product/market fit isn’t there yet and many of the conversations energy is spent on that. Clarity is important, but that doesn’t mean it’s there in the early stages. It depends on the startup’s situation. You can be clear but wrong too, but makes it even more problematic. But let’s not digress 🙂
After a career spent mostly in the $10-100M world, I can tell you with great assurance that product market fit is not an early stage problem exclusively.Companies hit a stall at $30M and more all the time and need to rethink everything including the product.
See my comment to Arnold below. The challenge is we’d all like to raise money from only from great partners. But the reality is that isn’t going to happen. The vast majority don’t raise money, and the majority raise it from sub-optimal partners where money truly is just that money. That’s just the reality. It’s the same at big companies too, I can tell you that as well, and not knowing you don’t know all the answers has bigger consequences.There is a very interesting tell: For most people as soon as you provide a response there is an instant response back. Or if you ponder their question for a moment, they either jump in and start talking, or ask you if you understood/heard the question.It’s taught in schools and many people assume intelligence equals being able to respond quickly with an answer. That’s not intelligence if it is a very clear situation, it’s really just memory. Intelligence is when you process the person’s question really think about it, have some questions yourself and be able to come up with the questions that need to eventually be answered and come plan to do that.No business is a clear situation ever, at the top level it’s coming up with questions that need to be answered and deciding how to work to the answer with the realization that just when you think you know the answer the question changes. Even things that seem clear like what does a preferred share mean has tons of nuances.That’s why I really have come to like writeups versus PowerPoints. Powerpoints. Could you imagine if all of these posts were PowerPoints not writeups? Generally Powerpoints are meant to try and show you know the answers. I’m not saying they don’t have their place they are great for communicating to large groups where there isn’t interaction.
Finally got to watching this toward my end of day, and it was very insightful. Very impressive how the partners clearly communicate so much information in few words. Pound for pound, or minute for minute, this was very packed with shop talk insights.If you put it on background, you’d miss half the points being discussed. So, I would disagree, and put it on foreground 🙂
i did exactly that — foreground.i’m enamoured by the idea that we are potentially headed towards more/all data being public.the real estate industry, we are basically the opposite of that today.
caught 10m of this and look forward to coming back…disqus bookmark
If you’re looking to unbundle the medical establishment from the outside, I’d encourage your medical-related start-ups to empower the alternative medicine field as well. There is such a disconnect between medical doctors and reality sometimes– not always!– but being constrained as they are by politics, pressure from insurance companies, drug companies, a lack of education on the connection between nutrition and health, how the body connects to the psyche, and the connection between our environment and our illnesses… alternative medicine is filling a huge gap. I noticed humandx doesn’t list naturopathic doctors on their list of acceptable users in their Terms and Agreements. Their training is accredited and just as rigorous, I believe (7 years).
This woman’s got her finger on the pulse of some technologies that are breaking the medical establishment open, so patients can be in charge of their own health, their own CEO’s if you will. http://www.forbes.com/sites…
Yup.Huge believer in functional medicine as it it encompasses both alternative as well as the most advanced technological and alternative treatments.In NYC one of the very best is the Center for Health and Healing in Flatiron.Amazing group of physicians.
It’s really nice to see that the main focus is the huge responsibility of being a leading tech investor: society, education, policy etc (the word “bubble” first appeared only in min. 39!). Tech is not just to beat the drums about the next new thing, but also to be thoughtful, even compassionate, about the impact on people and society. Great talk.