Video Of The Week: Chris, Marc, and Eric
Tyrone sent this video to me and I started watching it this morning on Chromecast in my family room. I don’t think I’ve seen Chris and Marc do anything together publicly. Since they are two of the smartest people I know, and since they work together now, it’s pretty neat to see them riff off each other. And Eric Ries is no slouch either. The entire video is about 45 minutes. I’m a third of the way through it and thoroughly enjoying it.
“The output of a technology company is innovation.” — Marc Andreessen.Brilliant. No mention of product / content / code language being the be-all-end-all in isolation.The techcos which have endured have innovated on multi-dimensions (product, code, usability etc) in parallel and simultaneously, that’s what I think.
In the long run every industry becomes a tech industry. No doubt. Interesting to see what different industries do with it. $CME has said it’s a tech company-yet while it invests a lot of money in technology, I don’t see it acting like a traditional tech company. (Lean startup, agile, charges big money for data, hierarchies, not flat etc). CME is also heavily regulated, and hides behind and helps write that regulation to put up competitive barriers.Regulated industries will be a tough challenge, since politicians that are bought could change regs on a whim. It also increases costs, which increases the cost of failing.
Funny–I think just the opposite that there are really no more tech companies at all.
Not classical tech-but any company that wants to scale needs tech.
As I wrote this I thought of a bunch of exceptions but decided to leave it be.Every company needs tech infrastructure just like they need marketing platforms. Just a piece of the operational puzzle.What’s interesting is that tech is the commodity piece now (Shopify et al). Marketing not at all so.
agree with you. Marketing is the most important major in a business school today. It used to be accounting and finance. When I started Hyde Park Angels (2007) in Chicago, people asked if we were going to be like “Silicon Valley” and fund tech companies. I said, “Every company we fund will have a technical component, but we are going to fund companies that build off the DNA of the city of Chicago and midwest”.If we decided to go head to head with the Valley and fund only companies that were Valley type tech companies, we’d fail.
True–Shopify if they get their shit together (which is not at all clear) can be truly huge.Marketing is hard to platform. The platforms are exploding but they are pandering to the wrong piece of their value.It’s fascinating to watch the platforming of these pieces. Lots to be learned from them.
Food and Apparel (When we speak of Tech today, we often really mean computer software.) There are other Technologies, like material science and food science that while not as mainstream still provide huge barriers to entry and are part of big markets.
had an engineer tell me the most valuable engineering major going forward is Material Science.
material science engineering, One of my Undergrad Majors.
You can still classify many companies by traditional labels. (Correct me on any of these because this is the way I see it).Intel – TechIBM – TechMacys – RetailNew York Times – MediaDelta – AirlineBoeing – Aviationhttp://lulitonix.com/ – Health, or Food etc.http://www.wanderandtrade.com/ – Ecommercehttps://www.riskalyze.com/ – Perhaps consultingUSV – Financial, InvestmentYour local wine retailer – RetailThere are of course businesses that are hard to classify by old labels but that doesn’t mean that there are no businesses that can’t still be classified by those old labels.
Categories are part of how we think.They are both useful to think through and negative to hold on to.Retail is how you sell not what. Add what and it starts to inform more clearly. Add the fact that most wine retailers do upwards of 20% of their business on line and out of area and retail itself is qualified.I would argue that from your list, tech is the most useless category as it tells you nothing and leads the conversation really nowhere. That’s why it as a term is really not that useful category wise.
How so?Being Intel is much different than being Macys. (Or Jon Malkovich for that matter).
So Walmart and Chambers Street Wines are both retailers.Tells me little.How about steak and sashimi are both food.They are about as useful as SIC codes which to me is not at all.Jon Malkovich though I may just rematch tonight–thanks.
Agree that it tells you little.But it does tell you something. That someone walks through the door of a retail location to buy products (if Walmart they look like a loser slob in many cases).This is quite different than the way that Boeing’s products are sold as well as the typical customers of Boeing’s products.So let’s do it this way. If you decided to sell your products at a “retail” location, that is a “retail” store as a general rule wouldn’t you pick someone with “retail” knowledge? As opposed to someone who sold chips for Intel? (Unless of course they were also a wine geek you get the point).Look I in no way think about these things like a business school professor a stock analyst might. Or my favorite – someone who writes about business!But even I find it a tad bit helpful to categorize certain things according to traditional labels if just as a shortcut.(This is a total Seinfeld we are having by the way. Because it’s so hazy I’m having a hard time of convincing myself of what I am trying to say and I almost never have that problem.)
True–but when categories tell you a little but not enough, they need to be rethought is my point.And retail, location based as one revenue model is really not relevant except in certain segments.
So your saying that the traditional categories are simply too broad to be useful in today’s environment ?
That and they have changed.Retail is now off and online, a million variations.Tech as a business model draws no picture in my mind.If they are not useful, find others.
The pendulum is swinging back to TECH.
Example pls of a company that ‘tech’ is useful and explains what they do, how they sell, and the value they bring.
What your definition of “TECH”. Biotech startups are tech first. For example, look at the portfolio of 5AM Ventures.
I am stating that the definition of tech is useless. I don’t have one.
Most hard tech goes unnoticed. Intel is a great example of the lengths a tech company has to go before it becomes a part of main street lexicon.
True–my point, that with some exceptions to most people except inside the VC industry, not a useful way to think.
Yes, in practice there’s no hope for a gooddefinition of ‘tech’. One major reason is thatit is in the interest of many players to corruptany clear definition.Still, for STEM, we are not much in doubt whatwe are talking about, including for the ‘T’ oftechnology.
Really interesting to hear Chris and Marc talk about customer development. I associate them mostly with big bets on industry changing companies, however as they discuss, to change an industry, you still need to identify a discrete customer segment, solve their problems, determine how you can acquire them, and build up from there. Wealthfront was a great example.
Chinese steelmakers dominate the US Steel industry. Their first target niche was rebar-which no one like to make, wasn’t sexy and had low profit margin. They took that market, and then blew up the silo.
Similar to Sam Walton choosing small towns for his stores which were being ignored at the time.
Consider that Chris and Marc do a ton of angel investing on top of their A16Z Growth Fund big bets.
Useful video, thanks for posting. Good to hear the comments on Business Plans as a way to more rigorously think things out. Ending comments about future trends including Crowd Funding and Big Data show me how much even these two very smart guys may be in the dark about future events, Black Swan situations etc. We don’t know what we don’t know, however clever we think ourselves. Liked the discussion about entrepreneurs in regulated industries becoming more and more “out there” by their fourth or fifth start-up in the regulated space. Funny image.
there some other great ones on that channel too.
Thank you Tyrone! Much appreciated.
“The art of leadership is when people follow you, if only out of curiosity.”That’s a great quote from Colin Powell. I think he aptly described the psychology of social media following.
Yes, I am not sure that is a good defining statement for the art of leadership. I like Schofield’s graduating address to West Point in 1879 better.
OK !You led me by curiosity to:Schofield’s QuoteThe discipline which makes the soldiers of a free country reliable in battle is not to be gained by harsh or tyrannical treatment. On the contrary, such treatment is far more likely to destroy than to make an army. It is possible to impart instruction and give commands in such a manner and such a tone of voice as to inspire in the soldier no feeling but an intense desire to obey, while the opposite manner and tone of voice cannot fail to excite strong resentment and a desire to disobey. The one mode or other of dealing with subordinates springs from a corresponding spirit in the breast of the commander. He who feels the respect which is due to others, cannot fail to inspire in them respect for himself, while he who feels, and hence manifests disrespect towards others, especially his subordinates, cannot fail to inspire hatred against himself.
“If you have no hypothesis, all the science in the world won’t help you.” — Eric Ries.Interesting because hypothesis are ideas of how the world works / should work. The science is the way to empirically measure and prove the idea.Exceptional technologists like Alan Kay of Apple have hypothesis about tablets (see Dynabook in 1970s long before iPod was launched in market). At the time there was no science (no code, no touch or graphical interface, no measurement of clicks/touch) to validate this idea of tablets.So innovators CHANGE the science itself to enable the validation of their hypothesis. Kay created Smalltalk, the precursor language to Objective-C which now powers iOS.Therein the hypothesis and the scientific tools are symbiotic and mutually transformative.Combine technical innovators like Alan Kay and Steve Wozniak with an articulate sales and customer-centric supernova like Steve Jobs and you create the most valuable technology company in the world.
Exceptional technologists like Alan Kay of Apple have hypothesis about tablets (see Dynabook in 1970s long before iPod was launched in market). At the time there was no science (no code, no touch or graphical interface, no measurement of clicks/touch) to validate this idea of tablets.Lots of people have ideas.  I thought of many (off the top I could mention advertising at self serve gas pumps and apartment selection books) but so did others. Either the idea is ahead of it’s time or you simply can’t execute or you can execute and you don’t. Jobs of course had Newton which was ahead of it’s time. People wanted to go to the moon for how long?Plus there is the rule of a million people thinking things (in other words survivor bias) whereby something is going to stick if enough people are trying enough things.In short: Futuristic things this are not worth quitting your day job over. My loser ex brother in law wasted time, energy and money in the 80’s on an idea that was essentially putting a fridge in the living room so you didn’t have to get up to get something to drink. He had neither the ability or the resources to bring that to market. Years later of course some company made a reasonably successful product just like that. Does that make him a visionary? (Realize this is nothing like an Alan Kay but thought I would mention to illustrate .)
That’s why the adage “Ideas have no value. It’s all in the execution” arose.The execution involves applying the existing scientific tools and/or inventing scientific instruments to solve a problem, preferably a big hard problem.That invention may be patentable and patents are attractive to investors.That then provides access to resources to bring the idea’s execution to market.
The pendulum is changing on ideas. On one end of the pendulum are technical IDEAS, these ideas matter because the ideas typically require brilliance, education and technical experience and creativity (TESLA etc.). On the other end are IDEAS that don’t (or matter less), these are the ideas that require mostly just a little creativity and merely assemble exhausting technologies (Kick Starter etc.). We are beginning to see the swing back to the IDEAS matter stage.
Let’s explore what if it’s not a matter of ideas or execution having more weight in the pendulum.I’m constantly reminded of how Quantum Theory really changes the world.You see, Determinism led our species to think of a single outcome and a linear path to that outcome, e.g.Life = food + water + sunlight.Tech success = product founder + investment + customersWhen Probability got invented (man-made) it led our species to think of two possible outcomes: 0 or 1. Something doesn’t happen or something happens.This affects each step-wise component of the outcome postulated by Determininism, so:Tech success = product founder (risk they don’t code or can’t deal with accounting) + investors (risk they will get diluted or will flip too quick) + customers (risk they are fickle or will need a lot of on boarding)Now, we’ve used Probability for a few centuries in answering questions such as:* nature or nurture?* idea or execution?* is entrepreneurialism an art or a science?* tangible or intangible?* real world or virtual?* down vote (-1) or up vote (+1)We swing between these two extremes and the tools we develop have reflected Probability’s binomial properties:* 0’s and 1’s bits* cellular automata that gets used in social relationship edge calculations* Osgood semantic differentials (hot-cold)Anyway, along comes Quantum Theory (another man-made invention) and specifically Schrodinger’s Cat. The cat is both dead AND alive. There’s a parallel universe of outcomes not a single deterministic one that’s subsetted by Probability within the bounds of 0 and 1.If we apply Quantum Theory to the matter, meaning and value of ideas relative to execution………I’m of the mind they’re symbiotic and matter equally.The fact is we all think AND do.If all that the Great Philosophers did was think and they didn’t do (communicate in fora, write and publish papers, travel to convey their ideas) then their thoughts would have no matter except within their own minds.Conversely, if all our species did was do instead of also think then we would never have examined our conscience about whether it matters that XYZ dictators are building armies and doing all manner of atrocious things.We humans are awesome precisely because we think AND do about things that matter.
“Thinking” and “doing” are both reciprocal verbs – NO?Thinking is however much more discarnate, much more portable, much more contagious, much more frictionlessly recombinant.Ideas and their frictionlessly discarnate recombinance have been continuously amplified/accelerated in lock step with our advancing communications technologies.Under network conditions it is a matter of ever-less time before the right combination of ideas land on the right fertile entrepreneurial soil under the right environmental demand conditions to grow whole new creative realities.Yet when we crack open the entrepreneurial execution can of worms it too is endless chock-a-block full of creatively required, technology-accelerated, implementation ideas.So the whole thing seems like a self-refferencial two-mint-in-one affair ?Much like your avatar !
Thinking and doing are indeed reciprocal and iterative too.In chemistry they’d be called a reflux process.In a playground, the analogy for thinking and doing would be Double Dutch:* http://youtu.be/arNf5u3rJYg…Energy transmission flows between the two and each gives the other mass and impetus .
My avatar is because:(1.) I’m Chinese so Yin Yang is intrinsic in my heritage.(2.) I do things with my head, heart and soul.(3.) The blue is for blue-sky thinking. The blood-red is for passionate execution and the white is for purity of self.
That ideas are worthless, etc., is anold denigrating insult of entrepreneursby arrogant VCs who don’t mind lookingridiculous. E.g., the J. Doerr interviewathttp://www.avc.com/a_vc/201…with”Ideas are easy. Execution iseverything.”andthere are “plenty” of ideas.Sure, bad ideas are easy, and thenall that’s left is execution. Witha good idea, routine executioncan be sufficient for grand success,e.g., many US DoD projects.Clearly Doerr wants to ignore anythinglike the path to good ideas. To him an’idea’ is just a short description ofa ‘business idea’, e.g., Facebook forcat owners. An ‘idea’ like, say, GPSjust does not occur to him. He gotaway with this arrogance and ignorance because for a while hewas ambitious and lucky. Then hisluck ran out; his poor judgment, e.g.,about ‘global warming’, got him offthe path, and his LPs got concerned.Denigrating an entrepreneur might look like a good, opening negotiatingtactic. I learned that lesson in thefifth grade: Like many boys that age,my handwriting sucked. And girlsthat age are little master manipulators.So, when I sent a note to one suchgirl, she said my writing was impossibleto read. So, I sent a sequence of several notes increasingly clear until finally I sent a note with allcrystal clear block capitals, and herresponses were all the same. Shewasn’t looking at my handwriting and,instead, was having fun manipulatingand insulting me. Doerr wasn’t lookingat the ideas and, instead, was havingfun manipulating and insulting entrepreneurs.It’s a trick of fifth grade girls. Lookedat deeper than just the thickness of asheet of paper, it was the fifth gradegirl and Doerr who made the foolish, objective mistakes.Doerr can keep saying that ideas areworthless, and I will keep saying thatI can’t use an idiot on my Board.
Who are the VCs you’d say have more of the intellectual chops to value ideas as well as the operational chops to help shape founders’ execution?
I can’t imagine a VC taking serious technical workseriously. Problem sponsors at NSF, NIH, and DARPAdo so routinely, but VCs, gotta be looking for hen’steeth.For one reason, likely the LPs would be outraged ifany of their VCs applied any value to serioustechnical work. For a VC, serious technical workand a dime won’t cover a 10 cent cup of coffee; suchwork and $1 million won’t cover a 10 cent cup ofcoffee.Essentially all VCs deeply, profoundly, bitterlyhate and despise any suggestion that technical workbeyond, say, just some coding, should be takenseriously, and they don’t take routine codingseriously.One reason is, the VCs want to feel like thesmartest guys in the room if only because they havethe thick checkbooks, and, thus, resent anysuggestion that there might be something they don’tknow that could be important.That the VC ROI has on average sucked for 10 yearsdoesn’t change any minds. For serious technicalwork in a STEM field, e.g., math, to be takenseriously by VC, maybe an early prerequisite is someyachts. The successes of James Simons or AndrewViterbi so far just don’t cut any mustard with theVCs.It seems that the VCs are following what their LPswant, and the LPs are solidly from the field of’financial investments’ dominated by the traditionsof private equity and commercial banking and whereVC is just a small sideline. The traditions are oldand run deep: The basic idea, going way back, is tomake financial investments only in financial assets,i.e., collateral. The special point about VC isthat it is willing to use surrogates for financialmeasures. The standard surrogate is ‘traction’.From all I can see, VC is in trouble: Another 10years of low, average returns may totally sour theLPs. Then what will be left is VC from personalcheckbooks and/or ‘crowd funding’, etc.Of course the obvious path forward for VC is toborrow from NSF, NIH, and DARPA — don’t hold yourbreath for that — to get powerful, crucial, coretechnology to provide valuable solutions, with hightechnological barriers to entry, to big problems.Gee, VCs would have to read the business plansincluding the chapters on the core technology!Horrors! Bluntly, the VCs would have to be able todirect reviews done much like the research communitydoes and, then, have the review results be a seriouspart of project evaluation. Horrors!Likely so far the VCs rarely get into their in-boxprojects with a significant role for serious work inthe STEM fields; people able to do such serious workrarely pursue VC funded entrepreneurship.Of course, necessarily the VCs are looking for’another Google’; it appears that so far there areonly a few such each decade; so, that the VCs getonly a few project proposals with serious work inthe STEM fields need not be a good reason for theVCs to ignore such proposals.From material such as the video of this thread, theVCs keep looking for ‘patterns’ but usually in thewrong places, that is, relatively superficialplaces. So, right, a large network of engaged userswith a network effect as a means of growth andbarrier to entry can work. But serious STEM workfor powerful, valuable results with a hightechnological barrier to entry is a ‘pattern’ lefton the cutting room floor.That the VCs mostly want to write their first checksbased on a ‘demo’ and/or traction might not be sobad, but that such a VC would serve on the Board ofa company where serious STEM work would be importantfor the future of the company could be a disaster.There is a statement from Marc & Co. that there areonly about 15 projects a year worth a Series A. Myview is that in time there could be 1500 suchprojects a year if the VCs took serious STEM workseriously.For now the solution for ‘information technology'(IT) entrepreneurs is to take the opportunities thatthere are and not to fight obstacles that can becircumvented.So, a big opportunity is that can (1) plug togetherone heck of a server for about $1500 in parts, (2)in some locations in the US get, in a living room,for less than $150 a month, an Internet connectionwith a static IP address, rights to attach acommercial server, and 15 Mbps upload bandwidth, and(3) get a huge collection of the best infrastructuresoftware for free or nearly so (e.g., the MicrosoftBizSpark program). Marc did mention that it’s muchcheaper to start a company now. Here he’s correct.So, what’s left to do is a good ‘business idea’,maybe some solid STEM work, and typing in some code.Then, just as a Subchapter S or LLC, go live, getpublicity, users, ads, and revenue and grow. If canget users enough to keep a $1500 server and 15 Mbpsof upload bandwidth busy, then soon can have freecash in the business checking account greater than aVC Series A check.So, such a path is clean, indoor work, no heavylifting. Likely don’t have to fight the EPA,unions, regulatory agencies, goons, zoning boards,etc. Can build up quite a business checking accountbalance before getting on the radar of anyone whowants to cause trouble, e.g., the patent trolls,competitors, DDoS thugs, etc., and, then, with thecash can fight such obstacles.A good example remains the early days of Plenty ofFish: One guy, two old Dell servers, ads just fromGoogle, and $10 million a year in revenue.For such a business, what about a Series A? Well,then, for, maybe, $5 million, have to go from 100%ownership to 0% ownership with a chance of gettingback to 60% ownership after a four year vesting planand have to report to a Board of VCs whose’fiduciary responsibilities’ to their LPs tend toconflict strongly with the interests of the founder,etc. Why take such a Series A? I know; I know;for the inscrutable ‘business acumen’, Rolodex,’connections’, highly questionable ‘wisdom’ (e.g.,watch the video) of the VCs. Anyone who can doserious STEM work is at risk of laughing hard enoughto slap their thigh hard enough to break a femur.The opportunities look terrific. The obstacles?Don’t fight them and, instead, just circumvent them.Stay small until have a good pile of cash, and thengo for the big time in a more standard way.Heck, it’s a lot better than trying to run 10 pizzashops! Heck, think of the cash it takes to startjust one pizza shop! And in such a business, eachpizza needs one on one work by hand.In contrast, if can keep a $1500 server busy, thenit’s just sitting there sending out ads and makingmoney essentially on its own while the owner doesother things.To see how much of a business might get from aliving room, maybe with some space in a sparebedroom, and a suitable residential Internetconnection, let’s do some back of the envelopearithmetic. Yes, the crucial point is having a Website good enough to get the users, but assumingthat, let’s see what revenue some modest resourcesmight generate.We can get powerful Web servers for $1500 in partseach.A VC firm once wrote me that they would not considera Series A until my Web site had 100,000 uniquevisitors a month.Okay, suppose (1) the site sends Web pages of400,000 bits per page, (2) each page has on average5 ads, (3) each visit gets sent on average 6 pages,(4) get paid (from some KPCB Mary Meeker data) $2per 1000 ads displayed, (5) the 100,000 uniquesamount to 300,000 visits.Then would need an average upload bandwidth of400,000 * 300,000 * 6 / ( 3600 * 24 * 30 ) = 277,778bits per second (and now that’s tiny) and would getrevenue of5 * 6 * 2 * 300,000 / 1000 = 18,000dollars a month.If can get the usage to keep growing and can halffill on average 15 Mbps upload bandwidth 24 x 7,then send15 * 10**6 * 3600 * 24 * 30 / ( 2 * 400,000 ) = 48,600,000Web pages a month, serve15 * 10**6 * 3600 * 24 * 30 / ( 2 * 400,000 * 6 ) = 8,100,000visits a month, serve15 * 10**6 * 3600 * 24 * 30 / ( 2 * 400,000 * 6 * 3 ) = 2,700,000’uniques’ per month, get revenue of5 * 6 * 2 * 15 * 10**6 * 3600 * 24 * 30 / ( 2 * 400,000 * 1000 ) = 2,916,000dollars a month, and send a peak of15 * 10**6 / ( 400,000 ) = 37.500Web pages a second. One 8 core processor with aclock of 4.0 GHz in a mid tower case with a total of$1500 in parts might be able to do this!So, in resources, we’re talking a good residentialInternet connection and a computer from $1500 inparts and revenue of$2,916,000a month. Beats the heck out of the pizza shopbusiness.If need more such mid-tower cases, then as therevenue grows to$2,916,000a month, get some more computers for $1500 each andput them in a spare bedroom with a window unit A/C.Right: Might hire an electrician to upgrade thehouse circuit breaker box and run some 240 V andextra 120 V power to the spare bedroom.The main thing needed might be just a good businessidea and a good idea for some unique, powerful,crucial, core technology. Right, “ideas areworthless and execution is everything”. So, somehowwriting some routine software, plugging together amid-tower server, getting a static IP address anddomain name, going live, getting publicity, users,ads, and revenue are “everything”? And I want toreport to a Board of people who say such things?People who say such things are trying to concentrateon what they know and denigrate what they don’tunderstand. A tough guy approach to war might be tosuck it up and go over the top as in WWI. Athinking guy approach might be to use an infraredlaser from 10,000 feet to put a 500 pound bombthrough a window. The same dichotomyapplies also to commercial Internet projects.See a lot of projects with the ‘tough guy’ approachand nearly none based on anything with graduateschool prerequisites? Okay: Call that a problemthe flip side of which is an opportunity.I need to get some dinner, finish a little routineto do a simple SQL SELECT, just a simplemodification of some good, old code, and finish alittle routine to take some URLs from the SQL SELECTand put them in a simple HTML table. It’s gettingthere! Off to dinner: My local diner serves a nicestuffed sole dinner!
VCs openly say that they’re not for every founder and also that for each founder who does raise VC money, they have to due diligence the VC as much as the VC does them.There are a number of articles out there on Steve Jobs proving VCs wrong:* http://blogs.wsj.com/ventur…* http://www.zdnet.com/blog/f…Maybe it’s not the VCs that founders need to source as Steve Jobs’ example showed.VC Don Valentine apparently called Jobs a “renegade from the human race” (source: Wikipedia) and declined to invest but referred him to Mark Markkula.
I just read one of your links on Jobs.It will always be difficult to evaluate a projectrun by a guy like Jobs before he has a track recordof great success. As a result, once a guy like Jobshas been successful, likely a lot of VCs will wantto be sure they are not on record as having told him”No”.But some evaluations are relatively easy and routineto do; the research community, NSF, NIH, and DARPAdo such evaluations all the time.So, for nearly all concerned before the users orcustomers are totally thrilled like Jobs thrilledthem, we can see a project ‘framework’ that makesearly evaluation quite reasonable:(1) Pick a problem where the first good or a muchbetter solution will fairly clearly be accepted as a’must have’ widely enough to result in financialsuccess.(2) Do some applied research to get the desiredsolution. Implement the research in software andlock it up inside a server farm where outsiders haveessentially no chance of duplicating the work.(3) Do the rest, nearly all of which should beroutine.For evaluation, for (1) mostly need just a good viewof the ‘market’. The ideal case would be a cheap,safe, effective one pill cure for any cancer; herethere would be no doubt.For (2), can evaluate that just the way research isevaluated many times a day in the researchcommunities and NSF, NIH, and DARPA.For (3), that’s supposed to be routine.What about what Jobs did or SnapChat, Uber,PInterest, etc.? Too tough to evaluate. E.g., forUber, what about regulation and barrier to entry?In some cases, even with traction significant andgrowing rapidly, tough to evaluate — a lot ofpotential slips between good, initial traction and agood exit via an IPO.A large network of engaged users with a networkeffect as a barrier to entry? Tough to evaluate,especially for the issue of building that largenetwork, i.e., getting to ‘critical mass’. It wassuper tough to see that each of Twitter, Facebook,SnapChat, and PInterest would become a social ‘musthave’ that would lead to “a large network …”.E.g., what about Disqus? The world has been awashin forum software for decades, back to IBM, AOL,CompuServe, Yahoo Groups, some highly polishedexamples (e.g., used at eGullet). So, just whyDisqus? Tough to see.Tough to see early on that Jobs’s special ideas ofuser experience, product design, material fit andfinish, the approach to apps, artistic elegance,etc. would work. As in one of your links, Jobs wastrying to use hard/software technology to bring anew tool that would be central to the lives of thecommon man in the street; nice goal, but tough toevaluate. Instead Microsoft was trying to bringcomputing to desktops for ‘document preparation’ andwhatever else developers could make successful.For my (1)-(3) above, a problem for VCs is that theyand their LPs have too little in examples orunderstanding. So, to them (1)-(3) look like yetanother case in their in-box of flim-flam to getthem to make a bad investment.Thankfully for US national security, the US DoD ‘gotit’, caught on, and did often just brilliantly wellright away 70+ years ago and since. Of course, theyhad a big advantage: They were willing to listen tothe best STEM people and didn’t have a lot of MBAfinance guys around! Instead the VCs and their LPsare heavily MBA finance guys or got lucky with someroutine software for some ‘apps’ or some such andhave contempt for good STEM work.As usual, Darwin will have the last word.
Ideas are worthless.Insight is priceless.A Steve Jobs 1-2 ( i.e., a left jab-right cross of insight): people would rather touch a tablet, versus just type on it; several enabling touch based products were emerging (screens, gyros, maps).Insight requires top-down and bottom-up.Really good VCs are bottom-up focused because they are wildly over supplied with top-down ideas the go nowhere.
> Ideas are worthless.Not so fast. Not always! At one time GPS was justan idea on the back of an envelope at a lab a little north west of Laurel, MD.At one time the fast Fourier transform was just anidea from J. Tukey to R. Garwin at, as I recall,a US Presidential Science Advisors meeting.Tell the oil field geologists or the X-ray crystallographers that the idea of the fastFourier transform (FFT) is worthless.Tell the people whoroutinely save 5-15% of direct operating costsfor the airlines by doing careful scheduling thatG. Dantzig’s idea for the simplex algorithm wasworthless.Tell Rivest, Shamir, and Adlemanthat their idea for public key cryptosystems (RSA) is worthless.Go back a few decadesand tell car owners that the idea for hydraulic valve lifters is worthless.Tell the modern worldthat the Shockley, Bardeen, Brittain idea fora transistor was worthless.Tell much of themodern world that the C. Townes idea fora laser was worthless.Tell the world of theInternet that the ideas of L. Kleinrock onpacket communications were worthless.Tell nearly everyone who uses digital data thatthe R. Hamming ideas on error correctingcodes were worthless; also tell this toA. Viterbi, co-founder of Qualcomm.And try to tell me that the ideas forantibiotics are worthless — they savedmy life more than once.A problem of the business world is that an’idea’ is just a very short description ofa proposed product or service, a description such as might be given toa prospective customer or user. Stillsome of these ideas should be regardedas having value. A problem, and a flipside is an opportunity, is that that businessworld tends to miss out out on crucial, core, powerful, valuable, defensible ideas that can be thekey to a valuable business. Instead theMarc view is that ‘innovation’ is SnapChat,PInterest, Uber, etc.; there can be goodROI in such things but for such a business”never be between a VC and the door whenthe lock-up period is over”.
“If you have no hypothesis, all the science in the world won’t help you.” — Eric Ries.Is that to say that big-data cluster-analysis analytics will never be able to compete with human generated inductive hypothesis ?
This NYT article comments on how current empiricism (Big Data) competes with intuition (human-generated inductive hypothesis):* http://www.nytimes.com/2012…
I’d add that Big Data clustering and correlations are the tools we have today only because they’re what mathematics can provide us with (Group Sets and Probability).However, it’s entirely possible that someone will invent a form of mathematics that can measure human perceptions and intuitions which are involved in “human generated inductive hypothesis” processes.
> it’s entirely possible that someone will invent a form of mathematicsYup!
I should disclose a personal interest in this. You see, I don’t believe Probability & Stats (and therefore Big Data) is the panacea for how we measure and interpret data — especially the subjective, qualitative, organic data of human expression.So I coded a simple little dice game to test and prove my little hypothesis against Probability using its own assumption basis.Suppose we have 3 dice which we roll. They are independent of each other; the outcome of one has no effect on the others.We ask 100 ordinary people which of these outcomes looks the most probable:(A.) 1, 1, 1.(B.) 1, 2, 3.(C.) 3, 1, 2.The results of my poll showed that over 90% of people voted for (C.) 3, 1, 2.Yet Probability says that all three outcomes are equally probable. It’s 1/6 x 1/6 x 1/6.This in itself is the basis of a hypothesis that something (another form of maths? another language?) PRECEDES probability and logic in the way we think.That has implications for current models of our brain, our thinking, our intelligence and the simulated versions of those in the entities known as Artificial Intelligence and Machine Learning which runs Big Data.It gets too easily forgotten that Probability was invented to deal with the limited events of an inorganic object (dice) played between two organic entities, Blaise Pascal and Pierre de Fermat.There will be a form of maths yet which accords with Relativity Theory (no 0 or 1 outcome per Probability but simultaneous 0 and 1) as applicable to organic matter such as language and human emotions.Now………investors with the intelligence and operational chops to go source the person(s) who can code and develop a system based on that idea and support them all the way……….Would change the world.
> I should disclose a personal interest in this.You see, I don’t believe Probability & Stats is thepanacea for how we measure and interpret data —especially the subjective, qualitative, organic dataof human expressionOn “Panacea” you are correct; probability andstatistics are not a panacea for “… humanexpression”.But, still, for an accurate description of “…human expression”, probability and statistics remaincrucial tools. But a big issue is how these toolsare to be applied.A good first step is to take the usual, earlyteaching of these tools and to lower in importancemost of the actual math that is taught explicitlyand to totally get rid of a lot of the non-math thatis taught by example, suggestion, hint, etc.With the desk thusly largely cleared, we get tostart over. A good place is to roll back to A.Kolmogorov’s 1933 paperA. N. Kolmogorov, ‘Foundations of the Theory ofProbability, Second English Edition’, ChelseaPublishing Company, New York, 1956. Englishtranslation of the original German “Grundbegriffeder Wahrscheinlichkeitrechnung”, ‘Ergebnisse DerMathematik’, 1933.Don’t really have to read the original and, instead,can do well reading ‘graduate texts on probability’from the usual suspects, M. Loeve, J. Neveu, L.Breiman, K. Chung, etc. These texts are all basedon measure theory and presume a good background inpure math from an undergraduate degree and most of aMaster’s degree. In particular want quite a lot offacility with set theory, ‘real analysis’, andfunctional analysis.Next to no one in US ‘information technology’ orInternet entrepreneurship has this background or hasseen such work done. Actually, a course in graduateprobability is not popular in US pure math, and inthe other STEM fields the fairly high pure mathprerequisites make graduate probability remainunpopular. For whatever reasons, such probabilityis more popular in France and Russia.Central there is the idea of a ‘random variable’.The first thing to do is to set aside all common andintuitive ideas of ‘random’. Instead, go out andget a number, from nearly anything. Then, presto,can consider that number as the value of a randomvariable. If the number was from something’deterministic’, then, still, just fine. Inparticular, a deterministic process, say, the pathof a haseball, is still a Markov process.So, in practice, collect a lot of data. Now havevalues of a lot of random variables; each numbercollected is the value of one real valued randomvariable.Can also have random variables that take values notin the real numbers but also in nearly anything –complex numbers, discrete sets, etc.Measurements on what people do easily qualify asrandom variables.Can manipulate random variables with high generalityand get new random variables.Conceiving of some measurements that do not qualityas random variables takes some cleverness, e.g., theaxiom of choice.A lot is known about such random variables; some ofthe best that is known is amazing beyond belief, andwithout some solid mathematical proof nearly no onewould believe much of what is known.With some justified mathematical assumptions from areal problem, some of what is known about randomvariables, and some good manipulations of randomvariables, can get some good new information, e.g.,on what people do. People can’t escape the power ofKolmogorov’s random variables.Some parts of some of the intuitive view of ‘random’can reenter the picture but only if are careful.> The results of my poll showed that over 90% ofpeople voted for (C.) 3, 1, 2.Of course. People with their intuition commonly getthings wrong, and that fact is neither new nor deep.If you want to understand how human intuition works,good luck. Similarly for human intelligence. Fromall I can tell, so far we have next to no idea howhuman intelligence works. No, a lot of sales hypeaside, the human brain does not work like an old ornew IBM computer, even if IBM puts $2 billion and5000 people on such a project.> This in itself is the basis of a hypothesis thatsomething (another form of maths? another language?)PRECEDES probability and logic in the way we think.There’s not just one way humans “think”. Mightguess that humans have a lot of intuition. But itdid appear that with Galileo, Kepler, Newton, etc.there was a lot of impatience with the lack ofaccuracy from human ‘intuition’ and also a desirefor something more solid. So, we got mathematics,physical science, engineering, technology — STEM,etc. An example is what Kolmogorov did. Likelymostly in his life he was as intuitive, emotional,etc. as the next man, but with the methodology ofmathematics he was able to invent some tools thatcan be more powerful than intuition — so didNewton, Fourier, von Neumann, etc.Yes, the STEM fields are inventions of humanintelligence so that in some sense that intelligence”preceded” the STEM fields, but we can’t, thus,conclude that such intelligence is ‘primary’, more’fundamental’, more ‘powerful’, or beyondexplanation from the STEM fields.> That has implications for current models of ourbrain, our thinking, our intelligence and thesimulated versions of those in the entities known asArtificial Intelligence.I see no immediate “implications” if only becauseunderstanding “our thinking …”, that is, how itworks, in much detail is so far beyond us.Still, in some cases, with good work in mathematics,etc., we can make progress on getting good estimatesfor what humans will do.
Human intelligence is currently measured as IQ by Binet-Stanford tests or Mensa-type tests. They are biased towards logic, rational Descartian deduction, spatial reasoning (transposes of flag sequences for example) and some lateral reasoning of language (e.g., identifying odd one out from apples, bananas, cherries, cucumbers, pineapples, tomatoes).Neither IQ nor EQ tests, for example, our senses:(1.) taste, no one measures and assigns intelligence to the world’s top chefs.(2.) smell, e.g. no one measures how a “Nose” in a flavors company has the intelligence to distinguish between hundreds of different apple flavors.[I worked as a “Nose” as a teenager.](3.) touch, e.g. no one measures Lionel Messi’s touch on the ball and assigns an intelligence to him based on that.(4.) listening. Yes, we can measure frequency in hearing but not our an intelligence for modulations in languages and accents.(5.) visual perceptions. This is different from optics. If there was such a test for intelligence, the intelligence of Salvador Dali is different from Mondrian.These are a few examples of how there is much work to do in understanding natural human intelligence and crediting us with that intelligence and not simply some quant-based IQ score.
Last time I checked IQ it was just thefirst principal component in a simplefactor analysis. So, it’s not so clearjust what IQ means. Some of the historyof IQ is ugly.Still, we’re a long way from computer software that could do well on an IQtest.
Re “People with their intuition commonly getthings wrong”, the people whose intuition told them not to go near the Twin Towers on the morning of 9/11 would disagree with that comment.It’s a mystery why and how we humans think and feel the way we do. Scientific advances such as the discovery of DNA provide some glimpses, but we still have a way to go before we fully understand our own existence.The thing about absolute empiricism is that it leads to definitions of “right” and “wrong”.If intuition is not yet measurable, though, how can it be “wrong”?:*).
No, poor Eric should stick to vague remarks onmanagement of small scale, low quality softwaredevelopment and f’get about ‘science’!
Eric’s definitions of hypothesis, science and proof (traction) is the business version.It’s NOT how a mathematician or natural scientist would define hypothesis and proof.I’m a maths grad and worked in the chemical industry as a teenager in Product Development with PhDs so have some frames of reference about how relative Eric’s use of “hypothesis” and “science” are, and whether these definitions would stand up to scrutiny by the Wolf Prize committee.I also went to Mgmt Sch so know that trying to make business processes more scientific happened with Frederick Winslow Taylor, so Eric Ries is simply continuing along that tradition.Mgmt concepts and processes are pseudo-science and pseudo-science can be useful; particularly if it does provide some tools of definition and metrification — even if they’re primitive in comparison to Fourier, Schrodinger’s Wave equations and complex biological definitions etc.
> Eric’s definitions of hypothesis, science and proof (traction) is the business version.Ah, you are in an especially generous mood today!In such a mood, you’d be able to go to my father-in-law’s chicken house, with 40,000 chickens, shovelup what was on the floor, and declare it high qualityfertilizer! Actually, it was still full of weed seedsand, thus, would create a heck of a weed crop ifused on, say, soybeans or corn!As an applied mathematician as an MBA prof, itis possible to do some good applied math, andmaybe some good science, in managementscience. E.g., much of the origin of the problemP versus NP came from attempts to solvepractical problems in business via integerlinear programming.Some of the best work in applied statistics, e.g., non-parametric hypothesis testing andexperimental design and analysis of variance,are in business.Business has not totally ignored the fact thata good answer to ‘supply chain optimization’quickly becomes a problem in stochasticdynamic programming — at one time HP knew this and worked on it.Wall Street actually can get interested instochastic differential equations and partial differential equations.The oil industry is well sold on appliedmath: E.g., the fast Fourier transformis key to seismic exploration. No onewould be foolish enough to run an oilrefinery without some good work inoptimization, at least linear programmingand, more recently, non-linear programming.E.g., it has appeared that students ofProfessor C. Floudas at Princeton arequite welcome in and near Houston.
Marc: What turn us off—top-down market size slide and a slice of that market; do bottom up and show evidence to back that up (around 32th minute).Fred, would love your take on this in terms of seed vs Series A.We try very hard to get our startups (accelerator program going onto seed) to gather qualitative evidences to do this (custdev at our stage).Others would argue that nothing matters other than Traction (quantitative), but I suspect that’s from a Series A perspective.What do you look for in market sizing in either stage?
Right! They talked almost entirely about everythingbut what was crucial! So, they were blowing smoke!What they want is to talk a naive entrepreneur intomaking a bad business deal.Due to the VC and/or LP traditions, they want toinvest in a traditional way, a financial investmentinto a financial asset, that is, with collateral. Much of what’s different about VC investing is thatthey are willing to use surrogate measures for thefinancial asset, and the favorite such measure istraction.Traction and a lot of innovation on a really good daycan be maybe as valuable 75% of the value oftraction alone! That is, innovation and a dime won’tcover a 10 cent cup of coffee!How can we know such things? Because VCs will look carefully at traction but will ignoreanything at all significant in innovation — theywon’t read that chapter of the business planor even the business plan itself.
Great share Fred.Bottoms up thinking is my business mantra and was great to hear it held up by these guys. It separates the operators and those in the know from the pundits and the phrase makers quickly.A great, smart riff to listen to.
That top down versus bottom up was just posturingand won’t stand up to examination:The top down example was fine as far as it went;it actually is good to know that the ‘whole market’is so big that if get just 3% of it can do well. Marc,then, wanted to imply that the top down guy was claiming that getting the 3% would be easy, whichthe top down guy did not and need not claim. Still,knowing that getting 3% would be good is good toknow.Of course, then have to move to how the heck toget that 3%?Doing just bottom up is not good: The reason is thatbottom up is down in the small details very difficultto estimate. In strong contrast, often the estimateof the ‘whole market’ from the top down approachis fairly accurate although made up of many details.So, one approach for the bottom up is to try to argue that can get, say, 50% of a niche marketthat is 6% of the whole and, thus, get 3% of thewhole. Now knowing the size of the whole iscrucial and says that the 3% will be enough.So, have reduced the bottom up part to examiningthe niche.So, knowing the top down figure of the whole marketadmits easier work on the bottom up part.Marc was just posturing.More generally the market size can be seen asa problem in statistical estimation. So, see what data is available and try to construct an estimator, say, unbiased and minimum orsmall variance. And can use the modificationof ‘statistical decision theory’ where have somedollar costs associated with being wrong andget those, say, from the business plan spreadsheet. So, an estimate 2% too largemight have much different cost than anestimate 2% too small. So, might not wantan unbiased estimate but one with leastexpected cost. In addition, in some casescan use the fact that time is involved andthat more information will be availableover time. Then can use the old A. Waldsequential testing ideas, basically anapplication of dynamic programming(discrete time stochastic optimal control).Come on, Marc, let’s actually be ‘technical’, okay?
Outright rejection of the top-down number forgets that the mgmt consultancies went through the bottoms-up primary research process to arrive at those numbers.There’s also the paradox that we quote the projections of Mary Meeker, Pew, whatever Benedict Evans posts and those are aggregate top-down data too.Investors like to see the bottoms-up research because it shows initiative and that even early on you’re establishing customer communication channels. Communication channels that can then pipe into sales generation.Again, it’s not a “top-down OR bottoms-up” fight. Where primary research is difficult to get (eg regulated industries), the founder has few options but to rely on “top-down” market sizing. Where the primary research is easier to get (eg consumer apps), the founder can do “bottom-up”.Everything is relative.
Mostly agree. But for your> Outright rejection of the top-down number forgets that the mgmt consultancies went through the bottoms-up primary research process to arrive at those numbers.Sometimes, yes, but maybe not. Commonly it’s easyenough to get ‘macro’ economic data without goingthrough all the low level details. E.g., the governmentpublishes macro data on GDP, median income, employment, etc., but so far I’ve never directly provided any such data to a government economicsurvey effort!E.g., for a lot of macro economic data can do a ‘survey’where ask, say, a few hundred or a few thousand people.Then, if have done well selecting a ‘simple random sample’,can use just the weak law of large numbers to getan accurate estimate, even if what are estimating is thesum of 100 million detailed ‘bottom up’ contributions.> Again, it’s not a “top-down OR bottoms-up” fight.Right. So, Marc was having fun patronizing anddenigrating the efforts of entrepreneurs pitchingto his firm. Marc was being insulting and arrogant.Not good.So, Marc wrote the C++ code for Mosaic. Okay.Nice work, but no great accomplishment in technology.For what Marc did in technology, we should alsomention Tim Bernes-Lee, the the National Center for Supercomputing Applications (NCSA) at the University of Illinois, Jim Clark, ARPA-Net, etc.
@awaldstein:disqus love listening to content like this!
What are some examples of the “hybrid models of viral propagation for enterprise products” that Andreessen mentions at the end of the video. Is he referring to social/word-of-mouth types of campaigns on social networks (targeted at influencers) combined with traditional prospecting?
socialmarketanalytics.com might be one. specialized silo, could go viral on twitter.
I’d put Dropbox in that bucket. You share a Dropbox file with a partner at another company. He likes the service and starts using it to share files with his department. After a while, everyone in the company uses it and IT is forced to support it. Growth was highly viral early on. I believe Dropbox now has a team doing direct sales to companies.
“The Idea Maze”Great stuff.Really thought Chris did a great job of layering that concept on top of the motivation of the founder. Very articulate and completely apt in my opinion.Thanks for this sharing this.
Idea Maze–ain’t that the truth at times.
he wrote a blog post about it in August: http://cdixon.org/2013/08/0…
I remember.I’m a fan of Chris’s pov. Was a follower of Hunch from beginning to end. And impressed with the exit as he nailed the output but the input is still in the future.
Today was first time I heard the phrase idea maze. i am also fan of his pov. I thought what he said in the above video about clusters was also pretty interesting.
@fred — Tying this in with Marc Andreessen’s closing comments on “Big Data” and its dispersion across sectors, are you following IBM Watson’s new ecosystem initiatives for NY?”We have to INSTRUMENT the data.” — Mike Rhodin, IBM Watson.6 minutes on YouTube:* http://www.youtube.com/watc…
In a feeble attempt to summarizeThe presentation posits that the socially-mobile collection and mining of data with customizable purpose-driven realtime visually-instrumented feedback from big-data analytics will soon be the baked in foundational fabric of all modern business and consumer software processes and that IBM is building out a next generation cloud-platform layer-cake to make it so ? – Digram at 6:18- SELECTABLE BUSINESS-FUNCTION SUB-PROCESSES/Apps DELIVERY as a cloud-service-platform- SELECTABLE GENERIC BUSINESS-FUNCTION DELIVERY as a cloud-service-platform- SOFTWARE DEVELOPMENT WITH INTEGRATED INSTRUMENTED-ANALYTICS as a cloud-service-platform- INFRASTRUCTURE as a cloud-service-platformI wonder how long it will take before the top-layer of this cloud-service-platform layer-cake will deliver:”the socially-mobile collection, minding and customizable purpose-driven realtime instrumented big-data analytics” to the rest of us as the first VisiCalc-like end-user accessible construction kit. ?http://youtu.be/Lv-sY_z8MNs…
Thanks, I had to google VisiCalc because I’d never heard of that before.I’m going to suggest that, in the near future, we’re likely to see numbers which are not put into rows and columns on a spreadsheet, but rather as modular cell nodes on a Knowledge Graph.The developments in Neo4J lend itself to that.
Great video – thanks for sharing.
Eric Friedman there some other gems on that channel from the lean startup confReid Garrett Hoffman @ The Lean Startup Conferencehttp://www.youtube.com/watc…Well worth the 20 minutes.Robin Chase @ The Lean Startup Conferencehttp://www.youtube.com/watc…Really diving into the Sharing Economy space.see here for morehttps://www.youtube.com/use…
For some more videos in this world.90 minutes with Marc Andreessenhttp://video.pandodaily.com…Almost two hours with Chris Dixonhttp://www.youtube.com/watc…Over two hours with their partner Ben Horowitzhttp://www.youtube.com/watc…And of course Fred on the same show, Pando Monthlyhttp://www.youtube.com/watc…
Interesting discussion but Marc’s assertion that the car companies are not “technology companies” because they’re bad at consumer software (GPS, maps, dashboard UI, etc.), don’t iterate on the same cycle time as today’s SV software companies, and have to deal with mechanical things like internal combustion engines is wrong. His caricature of car tech as rickety, gurgling, etc. is dated by two, if not three, decades and is disrespectful to the 10s of thousands of engineers who have massively, and continuously, improved every dimension of car tech over the past 50 years. When we slight a large slower moving tech company in the software industry (IBM, SAP, Oracle, Microsoft, etc.) we don’t allege that they’re “not a technology company” and that the startups are – we just call them what they are: large incumbents with all the advantages and disadvantages that that brings. But no reasonable person would say: “they’re not technology companies”. What Tesla is doing to bring startup innovation to cars is fantastic but they’re building on the shoulders of others who came before them – just as every tech innovator does. The folks who came before them were very much technologists and innovators even if they’re located in Japan, Munich, Stuttgart or Detroit instead of SV.
Thanks for sharing this – extremely interesting and solid learning points