The Hard Thing About Hard Things

I bumped into Ben Horowitz on Monday and had the chance to talk to him about his book, The Hard Thing About Hard Things. I haven’t read it yet, but I downloaded it to my Kindle app after talking to him about it.

I saw this morning that Brad Feld has read it and thinks it is awesome. I am not surprised and although I don’t normally recommend books that I have yet to read, I think this one merits that.

I’ve seen Ben in action and his experience and approach is valuable to entrepreneurs and CEOs. If you are one, I think you should read it.


Comments (Archived):

  1. andyswan

    not recommending a book recommending a person. perfectly valid and ordering for my kindle thanks

  2. Jorge M. Torres

    It’s good to see successful founders writing for an audience that is as interested in learning how to build and scale as it is in learning how to start. Matt Blumberg’s book also comes to mind. There’s tons out there on starting, but not as much on building and scaling.

    1. William Mougayar

      That’s very true. Scaling and growth are a different beast than starting-up.

  3. LIAD

    Read it over the weekend, had to see if it lived up to the hyperbole.Was an enjoyable and easy read, full of anecdotes and behind the scenes stories and revelations. (Working with Marc seems a lot of fun).Work hard. Treat people well. Don’t lose your morality or integrity. Never give up.Irony is he has experienced business trials and tribulations which most of us can only ‘dream of’.Found myself thinking a lot through the book: ‘boy, i’d love to have those problems’

    1. LE

      Treat people well. Don’t lose your morality of integrity.In business all of that nice guy stuff is fine as long as you aren’t losing your shirt and you’ve got the appropriate breathing room and ability to do the right thing. When that is not the case it’s always everyman for himself. Apple treated my ex brother in law well after his life changing disability because they make a shitload of money. If they were under financial straits and not raking in billions they would not have done what they did.People have integrity because they can have integrity and can afford to do so. The ones that have integrity and put others first (and don’t have money) typically end up in the poor house because they can’t make the hard choices.Work hard, of course. But I’m not sure what “never give up” is supposed to mean in real life. [1] Of course there are times when you have to give up both in business and in relationships. Sound’s like very “get yourself a good lawyer” type advice.[1] Thinking of what happened to the Japanese in WW2.

      1. LIAD

        they were my takeaways not things he specifically laid out.don’t agree with what you say about only having integrity when things are going good. integrity only comes into play when it goes against self interest.

        1. LE

          don’t agree with what you sayThat’s because you haven’t been involved in the multitude of transactions that I have with both companies and individuals spanning a long period of time. A long period of time.That experience ranges from dealing with larger companies and also individuals.I was dealing recently with a handyman who I hired to do some painting and repairs. At the end of the job he claimed that I was mistaken about things that he was supposed to fix (which were all in an email). So I said “ok no problem I will give you $150 more just take care of those things for me”. I didn’t argue with him even though I was right (I make mistakes but not with that type of thing). Not only that but I paid him in full for the job including the extra $150 (was about $2000 total iirc) even though he was not finished (I had dealt with him before of course).Well you know now maybe 3 months later I am still (fucking) chasing him to complete the work. Of course I could have held back money but I decided to take a gamble knowing full well of the outcome (because I do experiments like that if the money is small).I could afford to be “mr integrity” because the amount is small and there is little downside risk to me.The handyman on the other hand is acting predictably in that since he has his money he has no motivation to come back. Because he has more business than he can handle. So chase him I do.Now multiply that experience by 5000 other situations that all fit the same pattern. Have you ever been jerked around by companies owing you $50,000 back when $50,000 was like $100,000? I have. They don’t have the money they don’t pay. They don’t sell their car and pay you. They tell you to fuck off (I’ve had people infer they will shoot me with a gun by the way). My guess is if they had the money they would have more integrity than that.

          1. awaldstein

            Nope.Of course it is easier to be humble when you are wealthy but the idea that integrity is related to success or experience is simply not true.@liad:disqus stick with your thought. It’s the right one.

          2. LE

            LIAD stick with your thought. It’s the right one.Hmm. “It’s the right one” seems pretty absolute to me.You know the saying. Ask 10 doctors get 10 opinions.Go out and extend credit to people for “large” amounts of money and see which of those people or companies end up paying and which don’t pay you? And then do a rough correlation with the ability to pay and see if that factors into their decision making and integrity. Do you think that people who have gone through bankruptcy don’t have integrity? (Maybe or maybe they just don’t have any money…)So yes there is for sure a correlation between integrity and the ability (financially) to do the right thing. A correlation doesn’t mean there aren’t cases where that is not the case. And it doesn’t mean that people with money always do the right thing either. But it’s easier to do the right thing if you are not living from hand to mouth.For the record I have never lived hand to mouth and have luckily always been able to do the right thing. But I am honest enough to say that if push came to shove all that integrity would go out the window and I would put myself first.

          3. awaldstein

            Integrity is an absolute right.Bigotry, racism, gender bias are absolute wrongs.There is no grey area here.Yes–I’m right on this one.How you or me might act if the world was reduced to a chapter in Lord of the Flies matters not at all to this truth.

          4. PhilipSugar

            I agree. There was a good article yesterday in the NYTimes about never lying. I always said the worst lie is the one you tell yourself because you don’t address the problem.Yes other people don’t have integrity. So what, same as my comment the other day, you can’t use an excuse, well the other guy is worse.

          5. awaldstein

            You don’t have to be an ass or a crook to win.And trust me, I know how to be tough, manipulate the environment but I simply don’t do things I don’t like myself for.I learned to cut deals from a master, Jack Tramiel. Tough as nails and smart as they come. I find that now I get the same thing done without the drama and distaste.

          6. sigmaalgebra

            Sorry, I agree with both you and LE:In practice, too often LE is right on target. I saw a company highly regarded,being very generous, ethical, etc., untilthey lost $16 billion in three years andthen acted like low level street crooksseemingly taking pride in lack of integrity.

          7. JamesHRH

            Money doesn’t change you, it magnifies you.Applies if you go from having none to having tons……or the other way.

          8. sigmaalgebra

            Nice. A keeper! And you didn’t getthis from JLM? Ah, AVC is improving!

          9. JamesHRH

            Compliments always appreciated, Siggy!Not from JLM. I cannot claim authorship nor can I provide attribution. I forget who first said it to me, but I never forgot it.

      2. Donna Brewington White

        What does integrity mean to you, LE?

        1. LE

          Someone you can trust. But that’s fluid I don’t think it’s fixed on the person but more the person and the situation. For example you can trust me but the large corporation that I am dealing with maybe can’t trust me. It’s kind of a bespoke thing to me. I am always super wary of people who try to say they are super honest and treat everyone the same.Some people you trust that you won’t get ripped off to much since in some business situations you are sure to get taken for a ride you only hope to be treated reasonably fairly.So if you take your car into the local service station in Malibu to get fixed and the guy says “$175” so maybe he is padding the bill by $25 and maybe not. You just hope he isn’t going to try and get $500 from you. He’s padding the bill because in that business “you can only be as honest as the competition” and if he were 100% fair and honest the world would not beat a path to his door he would go out of business. Are there exceptions? Of course.

          1. Donna Brewington White

            Thanks for responding LE. My understanding of integrity is that it has more to do with the type of person you are than the situation that you’re in. You are absolutely right that it’s easier to be generous or benevolent when it doesn’t cause any pain or inconvenience to oneself or when it benefits you in some way. Integrity has to do with doing the right thing regardless of the situation or the personal consequences. Integrity is saying what you mean, meaning what you say, walking the talk, doing the right thing when no one’s looking. All that. It’s a very sad world or existence where this doesn’t happen. Happy to say that I know numerous people with integrity and who run their businesses with integrity.The gas station example… you can run an honest business, set a fair price and make a profit. Do you not think so?

    2. fredwilson

      Be careful what you wish for!

      1. LIAD

        lol.dealing with a crashing share price is predicated on first taking a company public.3 figure layoffs is predicated on first having a large companyyou get my point.

        1. Matt A. Myers

          Those would be all be predicated by a ton of smaller successes – with an incredible amount of factors that must come together to reach that point.. Persistence.

          1. Emily Merkle

            Amen. My thoughts exactly. We just launched a venture this week, and it’s taken the past 13 years to bring us to this very point. Cheap birth announcement:

          2. Matt A. Myers

            Congrats on the announcement. I was going to be doing an announcement 3 days from now — but that’s being pushed to May 1st now. Which I’m laughing at really in retrospect to how much time I’ve spent to get to this point. It’s a nice welcome buffer still. 🙂

          3. Emily Merkle

            Hey, thanks. The great unknown is so much fun. And exhausting.Looking forward to your news…

          4. Matt A. Myers

            I’ve been having a fantastic day too. It can’t get any better. 🙂

          5. Donna Brewington White

            Congrats, Emily. The great unknown gives us a reason to wake up each day. Happy to hear that you have given birth. Now comes the fun part, right?

          6. Emily Merkle

            Fun is hard work. But I wouldn’t have it any other way. Thanks, Donna.

          7. Emily Merkle

            Thanks, Donna. And it truly is fun. Even while living on lentils.

          8. Donna Brewington White

            May 1 is really soon. Wow, can’t believe you’re at this point. Exciting. Skype soon?

          9. Matt A. Myers

            My day/night got even better – rode the wave of good energy! ;)May 1st feels far away to me, but soon too. Lots to get done, but enough time to get it done. :)Skype – yes! 🙂

  4. Elia Freedman

    One of the great stories I read recently was his not hiring the ex-CFO, which helped him keep his company from being embroiled in an options backdating scandal. He is a great storyteller:

    1. PhilipSugar

      Actually I hate that one…Read the settlement: Here are excerpts. This was not simply granting at the lowest point of the month. To summarize it was not reporting, it was inflating sales numbers, it was moving money around oveseas, making up board minutes, lying on tax returns, continuing to backdate options after it was voted and agreed by the shareholders to have a specific standard and much more.I am sure it started off with picking the lowest price of the month. That is not where it ended.

      1. LE

        Actually I hate that one…Agree (but probably for different reasons).I think the much of the story is monday morning hyperbole such as:In retrospect, the only thing that kept me out of jail was some good luck and an outstanding General Counsel, and the right organizational design.So we are to believe that something was so obviously wrong was adverted because of an outstanding general counsel and that the other companies had average general counsels [1] and otherwise he would have “gone to jail”. Like really “gone to jail”.Hey guess what if you are running a business you’d better have a better seat of the pants feel of when something will land you in jail than that.Oh yeah one other things. This pretty much dovetails with what I said before on integrity:Still, firing somebody who had done nothing wrong at Opsware was tough. Nonetheless, Michelle graciously resigned as she did not want to bring negative attention to the company.Like really that’s actually what happened and the reason why it happened.[1] iirc didn’t many companies have the same issue and get into hot water?

        1. PhilipSugar

          Read my above post. That is where I agree somebody with no context and perspective reads and sits there shaking their head like a bobble-head.I read it and said: Huh?? Really?? PWC says picking the low day of the month is ok, your lawyer says not super clear after reading six times and you go to jail??I mean a month seems as aggressive as you could go, and is pushing the boundaries, but I mean do you pick the price the second you issue??? No, so where do you pick it? You couldn’t argue with the day, but the high or the low?. What about 5 business days? I mean people aren’t in holidays and stuff, you might have a strange bump?Didn’t seem right to me. So I looked what she plead guilty to.There is a lesson to be learned but it is not the one he gave.

          1. LE

            And to me it just didn’t feel right not even know what you know (this option thing isn’t “my thing” for sure) something just set off bells.Anyway not to load on Ben but you know I always try to figure out the basis for why someone does what they do or acts the way they do.My first gut reaction was to try to say that Ben’s parents were not smart because I’ve seen many examples of people using weak arguments (or oddly being con men) having come from families where they are smarter than their parents. [1] Anyway I found that that was definitely not the case as Ben’s father is quite the intellectual.But then I found this (because you know like any good defense attorney if you look under enough rocks you can find something to support your case, right?) about his father:…(See controversy and criticism section, errors, etc.)Acknowledging that this is a shitty thing for me to bring up but hopefully Ben won’t mind (I know I wouldn’t if I had his money..)[1] Reason being is that their parents never held their feet to the flame or called them on anything so they assume all the world is that “stupid” (I’m serious about that btw).

          2. PhilipSugar

            I don’t hold anybody’s parents for or against them, my wife’s parents abandoned her at 12. She grew up finding a place to stay through the kindness of friends.There was a good lesson to be learned here, but it was not here is why I didn’t go to jail. It was that you start off crossing the line and pretty soon its mighty far behind you.

          3. LE

            It was that you start off crossing the lineTotally agree with that. You start off small and get desensitized as you take on more and more risky behavior. Things progress. I have a related theory which I call “red light”. You don’t run the red light when nobody is looking late at night because you need your brain to recognize that a red light means stop. And if you run enough of them when you can “get away” with it you will inevitably run one when you shouldn’t (auto pilot action or whatever). Next thing you know tbone.I don’t hold anybody’s parents for or against them, my wife’s parents abandoned her at 12.My wife had an (almost) equally disturbing upbringing. I would venture to say (seeing you as a straight hardworking shooter) that your wife’s upbringing (or being abandoned) actually elevated you in her eyes. Meaning you can either be attracted to someone who is like your parents or the opposite.Anyway, my point is not about holding against someone for what their parents did or what they are. It’s more recognizing that the way someone’s parents are will (in many but not all cases) create the perspective on how they view the world (look at me vs. JLM as an example of that).So you learn things at an early age by the actions that your parents take (brainwashing). Doesn’t mean you can’t change and doesn’t mean you even know about all the things they do pro and con. But it’s there.(I would not pick you for this jury! Btw I’m not a lawyer.)

          4. PhilipSugar

            I think the difference in the level of difficulty of us completing an education at Penn is not even on the same plane and she received her masters. (I did not know this until after we were married and her grandmother brought up. She held an unbelievable grudge against her son, but my wife does not, we see her Dad all the time)

          5. Donna Brewington White

            Your wife sounds pretty amazing, Phil.

      2. Elia Freedman

        I’m not certain how that jibes with the story he tells. He decided not to implement her plan, and eventually let her go, because of the input from his lawyers that it doesn’t look like a reasonable plan. She went to jail for her part in the other company. Am I missing something?

        1. PhilipSugar

          His quote: “Michelle ultimately served 3½ months in jail for her part in the other company’s stock option practice—the same practice that we nearly implemented at Opsware.”Totally fucking wrong.She went to jail for some serious lying, and actually the jail came from purposely lying to the IRS about taxes. Here are some excerpts from the actual indictment she pleads to….full text below if you care to read, but here are some highlights :1. Mercury and the senior executives continued the backdating for years in spite of a specific change mandated and approved by shareholders in 1998 that required the exercise price of all employee options to be 100% of the fair market value of the company’s stock on the grant date.2. Backdating the grants by over four months3. In fact, Landan and Abrams at times backdated the exercise of backdated option grants. The company concealed from its shareholders the benefits reaped by these executives by making fraudulent proxy disclosures relating to officer stock option exercises4. Prepared false documentation memorializing the grants, including false written consents and meeting minutes5. Under-reporting income by $182mm to put options in the money: For example, a 1999 PowerPoint presentation by Abrams to Landan and others concerning the company’s financial picture stated in a slide: “Our Hidden Backlog . . . What Any Analyst Would Love to Get Their Hands On!”.The senior officers used hindsight to select the purported grant dates of the options, backdating the grants by over four months and making the grants in-the-money from 44 cents to $60 on the date they were actually approved. The complaint alleges that from 1997 through 2005, the accounting consequences of these benefits were then concealed as Landan, and at various times Abrams, Smith, Skaer and others, caused Mercury to fail to record over $258 million in compensation expenses and to provide false and misleading compensation disclosures to Mercury’s shareholders in filings with the Commission. Mercury and the senior executives continued the backdating for years in spite of a specific change mandated and approved by shareholders in 1998 that required the exercise price of all employee options to be 100% of the fair market value of the company’s stock on the grant date.The SEC alleges that the company backdated 45 different stock option grants to executives and employees, representing every grant made by the company to executives and employees during 1997 to April 2002. As alleged in the complaint, Skaer, or others at here direction, prepared false documentation memorializing the grants, including false written consents and meeting minutes. The complaint alleges that Landan, Abrams, Smith and Skaer each personally benefited by receiving backdated stock options that were in-the-money by, in the aggregate, millions of dollars through the fraudulent scheme.The complaint also alleges that from 1998 through 2001, Mercury, acting through Landan, Abrams and Skaer, fraudulently backdated the date of option exercises of certain senior Mercury officers. According to the complaint, senior executives were given preferential treatment and on multiple occasions were permitted to backdate the date of exercise of stock options with the company. The complaint alleges that these executives, including Landan and Abrams, backdated option exercises to dates consistent with low-points of the company’s stock, in order to minimize their taxable gain on exercise or receive more favorable long-term capital gains treatment on profits they earned upon the later sale of the stock acquired through exercise. For example, the complaint alleges that in connection with three backdated exercises, Landan was able to underreport over $18 million in gains upon exercise. In fact, Landan and Abrams at times backdated the exercise of backdated option grants. The company concealed from its shareholders the benefits reaped by these executives by making fraudulent proxy disclosures relating to officer stock option exercises, while Landan, Abrams and Skaer also concealed the backdated exercises in Forms 4 filed with the Commission.In addition, the complaint alleges that during at least 1997 through 2001, Mercury, through Landan, Abrams and others, secretly managed the company’s reported earnings per share (“EPS”) to meet or exceed financial analyst expectations by manipulating the recognition of revenue and making fraudulent disclosures concerning its sales orders. According to the complaint, Mercury stopped the shipment of its products once revenue targets for a period had been achieved, pushing the recognition of the revenue into subsequent periods. Between 1998 and 2001, this practice allowed the company to shift material amounts of revenue between reporting periods (from between $35 million to approximately $182 million in revenues). The company concealed the effect of this stop-shipment practice from the public through fraudulent and misleading statements and omissions concerning the “backlog” of its product bookings. Landan and Abrams understood that the backlog of revenues was material information that was being concealed from analysts and investors. For example, a 1999 PowerPoint presentation by Abrams to Landan and others concerning the company’s financial picture stated in a slide: “Our Hidden Backlog . . . What Any Analyst Would Love to Get Their Hands On!”Finally, the complaint alleges that during 1999 through 2005, at various times Abrams, Skaer, and others participated in the fraudulent structuring of loans for stock option exercises by overseas employees of the company in order to conceal the variable accounting consequences of those transactions, causing the company to fail to report approximately $24 million in required compensation expenses, which materially overstated the company’s reported pre-tax earnings during this period.SAN JOSE, Calif. – The former CFO for Mercury Interactive Corporation (Mercury) pleaded guilty Monday toobstructing the due administration of the Internal Revenue Service, First Assistant United States Attorney David L.Anderson and Special Agent in Charge, IRS Criminal Investigation, Scott O’Briant announced.In pleading guilty, Sharlene P. Abrams admitted that between 1993 and October 2001, she served as the ChiefFinancial Officer of Mercury, which is located in Mountain View, Calif. In April 2002, her joint federal tax return forthe tax year 2001 was filed with a Form W-2 that included a 2001 taxable income that was calculated in part on anumber of her Mercury stock options being exercised on April 4, 2001. When her joint 2001 tax return was filed inApril 2002, she knew that on April 4, 2001, she had not yet told Mercury what options she was exercising. She alsoknew that on April 4, 2001, there had not yet been a written action by the Mercury Board of Directors that formallyapproved the executive loan to pay for the options. Abrams was aware that Mercury’s Stock Option Plan requirednotice of exercise and payment before a stock option could be considered exercised. She also knew that on April 4,2001, her stock option exercises had not met those requirements. By failing to meet the requirements of the Plan,Abrams endeavored to claim an exercise date on which the company’s stock price was lower than on later dates. Thiscaused her to file a tax return and amended tax return that understated her wages and caused the IRS to assess anincorrect amount of tax for 2001, thereby obtaining a tax benefit for which she was not entitled.Abrams, was charged in a superseding information on Sept. 13, 2010. She was charged with and plead guilty to onecount of corruptly endeavoring to obstruct the due administration of the Internal Revenue Service, in violation of Title26 U.S.C. §7212(a).According to the superseding Information, in addition to her duties as CFO of Mercury, at various times between 1993and November 2001, Abrams served as Secretary and Vice President of Finance and Administration. She had theadditional title of Securities Compliance Officer as early as May 30, 1996 and was appointed a member of the stockoption committee on July 15, 1999. Mercury granted stock options to its employees and executives. Options grantedunder the stock option plans could be either incentive stock options or non-qualified stock options.According to the Superseding Information, Abrams supervised the exercises of stock options for officers andexecutives. At times, officers and executives exercised their stock options and then held the stock for a period of time,usually at least one year, to take advantage of the lower long-term capital gains tax

  5. JimHirshfield

    Nice rankings:

  6. William Mougayar

    It’s Hard to Ignore a Hard Hitting Title like that.Apparently, Ben suggested that title himself, and the publisher accepted it, which is rare. Typically, publishers are hard nosed about titles and they’ll often make the author change them for marketing purposes.

    1. LE

      A total airport rack best seller for sure. Especially with the back of the book reviews starting off with one by Zuck.

      1. Timothy Meade

        In the vein of airport reads: Book by some guy with forward by THE FACEBOOK GUY! !

    2. Emily Merkle

      I am ignoring it. The Kindle version is overpriced. Wonder what the thinking was behind that.

      1. LIAD

        found it weird too.also was priced $14 in US store, but only £6 (which is $10) in UK store

        1. William Mougayar

          $17 Canadian on Kindle. We’re getting hosed.

          1. JimHirshfield

            Take off, hoser.

          2. Matt A. Myers

            Hosed is politely put..

  7. LE

    While it certainly seems like interesting business entertainment reading I think my issue is the title which once again narrowly defines (and that is what I feel is distasteful) what a “business” is. And gives younger people a perspective that the only thing worthy of doing is the type of startup that is essentially a moonshot.The Full title is:”The Hard Thing About Hard Things: Building a Business When There Are No Easy Answers”But it’s not relevant to business it’s relevant to the startup world that is known to Mark and Ben. That’s a very very small world. The things they have done are atypical.This reader review on Amazon (literally at this time only 1 of 2 negative reviews) says this:Horowitz writes for startup CEOs who raise $20 or $40 million from top VCs. His own history, recounted in the book, shows just how irrelevant his own experience is to 99.9% of entrepreneurs.This is a book for CEOs of VC-controlled startups. Its lessons are unlikely to apply to real businesses owned by–funded by–real entrepreneurs.What makes sense for High Finance Entrepreneurship is usually suicidal for 99% of real businesses. VCs make their money by funding and promoting aggressively, and they don’t care about losses. That’s a luxury Main Street entrepreneurs cannot afford.Finally, I would urge readers to be cautious of any advice given by a VC. It’s typically self-serving and cannot be trusted.I highly recommend spending time reading Michael Church’s essays on VCs and VC-backed startups.My background: I’ve been a VC and led two VC-backed software companies. Multiple exits, high IRRs.I make money from startups following their dreams of course so all of this is good for me!! But I would be remiss in not pointing out that this type of dream is not the right path for most people. And I fear many people not pursuing things that could be worthwhile and achievable by trying to be the next Zuck. So there it goes. Ride at your own risk. I said it. (And I know people will still continue to go on the ride so hey I did my part.)

    1. Matt A. Myers

      I haven’t read it, though switching “business” with “life” seems more encompassing, more inviting and welcoming to non-business folk. Another term for a business is an organization, and we all can use a little (or a lot of) help and advice when it comes to organizing our lives.Edit: To add, I don’t know if any of the advice would be fitting for general population, though I always find parallels to non-business situations in business-related texts.

      1. LE

        disqus hung on my reply to this. NB.

        1. Matt A. Myers


          1. Simone Brunozzi

            Cool. n.b. is both Latin and Italian, and in Italy we use it often. Hoever it goes with punctuation, like n.b., not as “NB” – at least in Italy.

          2. Matt A. Myers

            LE is an innovator. And he clearly likes 2 letter combinations.

          3. ShanaC

            note bene!

    2. PhilipSugar

      You can learn many things even though you are not in that situation.

      1. LE

        That assumes 2 things.1. That the time couldn’t be spent better reading something else (time is not infinite).2. That the person reading has some context that allows them to put the lessons in perspective. My point is that many people who read this book have no frame of reference. They are open slates. You are not the same guy that you were when you graduated Wharton . You now have context. The guy|gal graduating now does not. No matter how bright. They will read in a different way. (Could be a plus could be a negative (see wanamaker).)Another example might be if I read a book on owning a private plane vs. JLM. JLM has context and I don’t. He knows which chapters/lessons make sense and are right (from his perspective) and can add to his knowledge base and which aren’t. I don’t. It goes back to that John Wanamaker truism “I know I waste 50% on my advertising I just don’t know which 50%”.

      2. awaldstein

        I’m with you and @aaronklein:disqus on this.Context is in your need, not in the specifics of the situation itself.I’m looking forward to this book.

      3. LE

        My opening paragraph was:While it certainly seems like interesting business entertainment reading I think my issue is the title which once again narrowly defines (and that is what I feel is distasteful) what a “business” is. And gives younger people a perspective that the only thing worthy of doing is the type of startup that is essentially a moonshot.Note the bolded part.Of course you can always learn something. I learn things from making comments here (and people’s almost predictable reactions and emotion) as well as having conversations with random people to see what makes them tick.You know you should go to Wharton and ask them if you can be a lecturer in entrepreneurship. You know what they will tell you they will say “you don’t have a Phd so we aren’t interested”. What they really mean is “you are ‘just an entrepreneur’ and “nobody knows who you are” so thanks but no thanks”. When both of us know there is so much that someone could learn from you at Wharton.

        1. PhilipSugar

          Actually, I speak twice a year at Wharton and UofD. Since I think this thread is over, I’ll give you the resume of my friend who is a Senior Fellow. He like me has two undergrad degrees.Rob Weber is Senior Fellow for the Fisher Program in Management and Technology and a Managing Director of Antiphony Partners, LLC, an innovation strategy consulting firm. Prior to Antiphony, he served as President of Intellifit Corporation a provider of body scanning technology for multichannel apparel applications. His entire career has focused around starting and building technology based enterprises, including operational roles as President of Elastomeric Technologies, a specialty electronic connector manufacturer, Marketing Director for Ensoniq Corp, a leading manufacturer of electronic musical instruments and multimedia products and President of knoa, an e-learning software company. He has also served as Managing Director of the business consulting firm Antiphony where he focused on technological innovation strategies. Weber has invested in dozens of start-up technology ventures through angel venture groups Robin Hood Ventures and the Mid-Atlantic Angel Group Fund, both of which he co-founded. He is a graduate of the Jerome Fisher Program in Management at the University of Pennsylvania where he received a BSE from the Wharton School and BAS from the School of Engineering and Applied Sciences.

        2. PhilipSugar

          I should add if I wanted to do for a living and make money you are 100% correct.

    3. ShanaC

      Truer than I would like

    4. Aaron Klein

      Totally and completely disagree. Could not disagree more.Haven’t taken a dime of VC and won’t for Riskalyze. I have learned a HUGE amount from Ben and the book, which is an incredibly good read.

      1. LE

        I said “it’s relevant to the startup world that is known to Mark and Ben”That doesn’t mean that someone has to take VC money to be part of that world.Suffice to say if you go out into your town and survey the businesses that are operating there (I’ve dealt with these types of businesses extensively what would be called traditional businesses) the lessons from that book will not be relevant to the majority of those businesses in a way that they can be actually be implemented. Sure a good read. But probably more in a motivation seminar type of way. But it reminds me of when I was in business school and the things that they taught you there. “So you raise equity and then you hire a VP of …” wow it’s that simple. Just do it.The lessons learned from this book (and once again I haven’t read it to be fair) are based on what was learned almost certainly with a shitload of money that they had to start loudcloud (which failed by the way a failure that if not for the shitload of money or fame of the founder (you’re not him, right?) would not allow a pivot to something that succeeded). And the way you grow or the chances you take with that type of money is not (to me) relevant to what you do if it’s your money and ass on the line. Could there be things to learn? Sure, ok there are always some things to learn.By the way you have a cert issue with (see the attached image in case you are reading this on email…)

        1. Aaron Klein

          My only point: if you are a growing company with >10 employees, there are HUGE learnings in that book.PS: Thanks for the heads up! We never use that domain, but it’s supposed to cleanly redirect.

          1. LE

            We never use that domainWho’s “we” sucka? [1] Your customers are “we”.Of course you dont’ use it. People trying to get to your site who do a typo use it. For the life of me I couldn’t remember the correct spelling of your site. So luckily you have that typo (might have even been one of the ones that I suggested don’t recall..) which is what I typed in.[1] Dirty Harry.

  8. Eric Friedman

    Its honestly amazing. The entire “good product manager/bad product manager” could be written to any job function. It has already been helpful to me as well.

  9. PJ

    This is indeed one of the best books I have read. Went straight through it and caused me two sleep-depriving nights. Too bad there isn’t a Chinese version for me to share with my portfolio CEOs. So I had to translate the parts I wanted to share myself.

  10. Michael Bloom

    Fred, you’ll love it. Not only for the learnings and killer stories, but it’s a great trip down memory lane for those of us who were building Internet companies back in the early days of Netscape and the web.

  11. Duncan Logan

    Ben gave me a copy @SXSW. I gave it a casual glance back in my room and ended up still reading at 6am. I was gripped, I think it is exceptional. I think it has an humility and honesty which is extremely rare in a “this is my story” sort of book. Entrepreneurs who have lived the “if this then that” where “that” is game over will deeply relate to his story. (who hasn’t?) I think the advice around dealing with company politics, hiring, firing, culture etc is fantastic, no matter how big or small your company. Seriously, I have ordered copies to hand out. I think it is that good.

  12. falicon

    It’s available on as well (… ) …been listening to it the past few days and have to say it is FUCKING awesome. He doesn’t just give advice or thoughts, he tells the real stories behind what shaped his scars and his thinking…read it.

    1. Matt A. Myers

      Ah. I always forget about Audible / audio books. I’d get a lot more “reading” done this way …

      1. falicon

        Yes – it’s a great way to get through some really great books!

    2. Timothy Meade

      Cool. It’ll be next in my queue.

  13. pointsnfigures

    someone needs to invent a memory chip I stick in my brain. so many books, so little time.

    1. Emily Merkle

      You don’t have one yet?

    2. LE

      so many books, so little time.One of the reasons to read books is enjoyment. Another is to learn something. To me this book is 85% for enjoyment and 15% to learn something (pick your own arbitrary numbers).Kind of like do you get pleasure from the food in your stomach or rather the process of eating the food?I’m not a big believer in what can be learned from “books like this” (this is w/o respect to this book which I haven’t read so yes I make assumptions as far as the content).While I’ve run into many people who have told me stories of things they have learned in direct contact with people (because that’s interactive and filled with nuance for a specific situation) and I know many people (including myself) who quote things and lessons from books and movies, I really haven’t anecdotally picked up on many people who have said “so I read this book by such and such when I was young and that is what helped me do x y and z later in my successful life”.That said (taking the other side) I’m a big believer in “the thing that leads to the thing” so it’s quite possible that people have been influenced by something but attribute their success to something other than a book that they read. In all fairness.

      1. sigmaalgebra

        > “so I read this book by such and such when I wasyoung and that is what helped me do x y and z laterin my successful life”.In my first effort at math grad school, thedepartment and I had very different goals andapproaches. So, I got a job (quite nice, actually)and (being single) got some math books and dug in,slowly, carefully.I started with linear algebra. Since my ugrad workhad had a course in abstract algebra, I knew some oflinear algebra. I got more from the notoriouslydifficult Nickerson, Spencer, and Steenrod,”Advanced Calculus’. With a reading course intheoretical physics, I learned some more. With myugrad math honors paper in group representations(all in terns of linear algebra), I learned more.Read through the first half of von Neumann’s’Quantum Mechanics’ — basically Hilbert spacetheory, basically linear algebra but in infinitelymany dimensions instead of just finitely many. ThenI got a relatively difficult book and worked thorughit and then got the classic, Halmos, ‘FiniteDimensional Vector Spaces’ and worked very carefullythrough that. In my career, I did more in linearalgebra — numerical methods, software,applications, etc.In my second attempt at math grad school, there wasan advanced course in linear algebra. I looked atthe content and told the faculty I didn’t need it.They smiled and said, “Then, take it anyway.”.The homework was long and carefully graded. Oncethe grader made a mistake on my paper; I correctedhim; and he made no more mistakes! When the coursewas nearly over, the grader asked me if I knew how Iwas doing. I didn’t, regarded material as quiteroutine and nearly all beneith me, and guessed I wasdoing okay. He showed me the scores across theclass, and I was far ahead in homework, tests, andmidterm. Net, I blew away the other students. Thereason? Mostly that book by Halmos. Who wasHalmos? An assistant to von Neumann at theInstitute for Advanced Study. His book was a finitedimensional version of von Neumann’s Hilbert spacework. Nice book.On a job, my manager was struggling with inverting amatrix involved in imaging laser radiation throughslits. The components of his matrix were notnumerical but algebraic. But, what I’d learned inlinear algebra worked fine, and I found his inversequickly. He was amazed.Once I was on a job interview, and there were twoissues the group was struggling with. One of thesewas some really bad numerical results frompolynomial curve fitting. But I’d looked at a bookon numerical analysis, as I recall, by Forsythe, andseen that a better approach was based construcingsome polynomials orthogonal for the data inquestion. That solved the problem, and I got thejob. I solved the second problem, also, from somematerial in Knuth’s ‘The Art of ComputerProgramming’ — here I needed to be tricky, overlayan array of 32 bit floating point numbers with twoarrays of 16 bit integers and then clean upafterwards, but my code worked.Elsewhere in this thread I mentioned doing some workon the US SSBN fleet. Well, the key book was E.Cinlar’s book on stochastic processes.For my current project, there are some books I readcarefully that are key as prerequisites for theoriginal work I did.In some work in Monte Carlo, a group was using arandom number generator of questionable quality. Igot the right volume of Knuth’s ‘The Art of ComputerProgramming’, looked up his recommended generators(that passed the Fourier test) wrote some assemblercode to handle (in an ‘object oriented’ way, i.e.,my code was a ‘method’ of a ‘class’, right, beforeC++, etc.) all such generators, settled onmultiplying by 5**15, adding 1, and taking modulo2**47, and gave it to the Monte Carlo work. Result:The old generator had been of low quality. Thankyou Knuth.Once I was challenged that some of my research wasnot ‘publishable’. I read Knuth’s ‘The TeXBook’,typed in my math, and got a highly polished paperthat published right away. Thank you Knuth.In my second try at math grad school, I needed onemore course for a MS, had seen a problem in a coursebut not solved in the course, so asked for a’reading course’ to attack the problem. In thecourse I didn’t need to do original work; expositorywould have been enough. So, I attacked the problem.I’d worked carefully, more than once, through W.Rudin’s, ‘Principles of Mathematical Analysis’ andremembered a construction, using an exponential,that yielded a function from the reals to the reals,zero at the orgin and on the negative half line,strictly positive otherwise, and infinitelydifferentiable. Using that and some other materialroughly connected with Rudin gave me a solution tothe problem. Along the way I discovered and proveda new theorem comparable with the classic Whitneyextension theorem. Word spread in the department,and then I had a nice, shiny halo. Later Ipublished the paper. Thank you W. Rudin.In some work for the Navy, we were working withpower spectra of ocean waves. Well, Bell Labs hadbeen interested in the power spectra of second orderstationary stochastic processes from which wasBlackman and Tukey, ‘The Measurement of PowerSpectra’. I got a copy and over a few days read itover dinner of broiled flounder, french fries, andbeer from small glasses.Soon our software house was bidding on a project andhad a specification from the Navy. I looked at thespecification and from Blackman and Tukey saw someproblems and wrote some software to illustrate thedifficulty. So, my software generated and passed asample path of a second order stationary stochasticprocess through a filter with a particular transferfunction, accumulated the empirical power spectrum,and, thus, showed how much data, as in Blackman andTukey, was needed for an accurate estimate of thepower spectrum. Thus I corrected the Navy’sspecification, and our software house gotessentially sole source. Thank you Blackman andTukey.And I have several more such examples.Net, books can help in business.

  14. awaldstein

    Downloaded and taking to the beach.There is lots to be learned from huge success. Those are the people who should be writing these books.

    1. fredwilson

      Tulum, right? Jealous

      1. awaldstein

        Yup..KIndle loaded. Wine coming with luggage. All is good.

        1. JimHirshfield

          Bon vinage.

        2. Donna Brewington White

          Enjoy, A.

  15. Elie Seidman

    highly recommended. I don’t know of anyone who does a better job ofcapturing what it feels like to be inside the ups and downs that a lot ofentrepreneurial ventures go through.

  16. paramendra

    I am trying to get hold of the USV phone number. And the infinite scrolling is rendering it impossible. Just when you hit bottom of the page, the bottom falls out.

    1. JimHirshfield

      It’s right there next to their fax number.

    2. fredwilson

      That’s a feature not a bug ;)I don’t think we publish our phone number on our website anymoreOur front office was getting inundated

      1. paramendra

        I managed to scroll down to the bottom of the page on my phone, take a screenshot, get the number. I did call. You are impossible to get through to. No news there! I am pitching. I have a startup finally.

  17. aminTorres

    Your “kindle app” is that on an ipad?

    1. Emily Merkle


  18. Twain Twain

    Peter Thiel’s ‘Zero to One: Notes on Startups, or How to Build the Future’ (shipping in Sept 2014) appears in the list of books associated with Ben Horowitz’s book.Its marketing pitch is: “The single most powerful pattern I have noticed is that successful people find value in unexpected places, and they do this by thinking about business from first principles instead of formulas. Ask not, what would Mark do? Ask:WHAT VALUABLE COMPANY IS NOBODY BUILDING?”If nobody’s building it yet…It’s probably really HARD to do.Re. Ben Horowitz’s book, it’s likely as invaluable and practical a read for startup CEOs as this blog post of his:*

    1. sigmaalgebra

      > If nobody’s building it yet…It’sprobably really HARD to do.Maybe not!How ‘hard’ something is to do can depend alot on what tools, etc, are brought to thechallenge. Something that is really’hard’ for nearly everyone might be quiteeasy for someone bringing better tools tothe challenge.Here’s a challenge: Live in a desertcountry; tend sheep at night; watch thestars in the usually astoundingly clearnight sky; see all the stars essentiallyfixed with respect to each other; observea few stars that ‘wander’; and figure outwhat the heck is going on. Some verybright; hard working people struggled withthis question for many centuries.Finally Newton got a very clean solutionfrom his second law of motion and law ofgravity. How? “If I’ve see farther thanothers it is because I stood on theshoulders of giants.”In WWII; the forces under Nimitz andMacArthur too often suffered unspeakableagonies of the damned. A prediction wasthat an invasion of the home islands ofJapan would result in 1 million UScasualties. Finally Oppenheimer’s workgot Japan to surrender in; what, twoweeks? Oppenheimer brought a lot more tothe problem. Oppenheimer’s work costabout $3 billion, that is, about $3000 percasualty. A great bargain.At about 10 I had a tummy ache. Mom wassmart enough to call a doctor and had beensmart enough to know one of the bestgeneral surgeons around. We met at theentrance to the operating room, and thesurgeon gave me an exploratory. Myappendix was inflamed; I had an infectionof my peritoneum and illium. He removedthe appendix, sewed me up, and for twoweeks in the hospital filled me withpenicillin, streptomycin, and terramycin.I walked out well. 100 years ago, areally ‘hard’ problem; when I got sick, arelatively easy problem.Once I heard about a project in troubleand got involved. The promise was that Iwould get a good job! The problem was inallocating marketing resources to asudden, new business opportunity. Anoptimal solution would be of interest.The problem had been formulated as 0-1integer linear programming. At the time,the ‘fad’ technique for a solution was’simulated annealing’. So, the team hadtried that, run the software for days, andstill didn’t know what the heck they had.I’d studied optimization in one of thebest graduate programs in the world andknew some nonlinear duality theory andLagrangian relaxation. The 0-1 integerlinear programming problem was 40,000constraints and 600,000 variables. So,one week I wrote out some math and asecond week, some corresponding software.So, my software executed Lagrangianrelaxation iterations for 905 seconds on a90 MHz PC. From the bounds from theduality, I saw that the best feasiblesolution I had had to be within 0.025% ofoptimality. Right: In general 0-1integer linear programming is in the setNP-complete! So, I beat the pants off the’simulated annealing’. Why? I broughtsome good tools to the challenge. ‘Hard’?Without the tools, yes. With the tools,not really.I didn’t get the job! I’d been very nicethe whole time, but my work had been somuch better than the team’s that the teamwas humiliated and afraid and angry withme.Once the US Navy wondered if the US SSBNfleet could be vulnerable to a special,controversial scenario of global nuclearwar but limited to sea. Here’s the point:If are in nuclear war and find asubmarine, then killing it is easy. Butif are not in war and kill one or just afew SSBNs, then the rest will fire and warwill be started and over. But if are notin a war and want to find all the US SSBNsat the same time and kill all of them atthe same time, ‘rots uh ‘ruck, and therewon’t be any!But suppose a war were limited to sea andextended over some days or weeks. Thenwould the SSBN fleet be vulnerable tobeing picked off one by one at asignificantly rapid rate! Hmm?Uh, the Navy wanted their answer in twoweeks! I was supporting my wife and methrough our Ph.D. degrees and was asked tocontribute. For tools, I brought to theproblem knowledge of some old work by B.Koopman and continuous time, discretestate space Markov processes. For theproblem I saw such a Markov processsubordinated to a Poisson process, wrotethe software to do an evaluation, and theNavy got their results on time. ‘Hard’?With the tools I had, no. I thank E.Cinlar, now long at Princeton, and one ofhis star students. But, then, I learnedthat there were other teams also workingon the same problem, in parallel, at thesame time. What those teams were tryingto do was ‘hard’, especially in two weeks.None of those teams had results in the twoweeks. My work? Fast, fun, and easy.The other teams and their managers becameafraid of me and angry!Something can seem ‘hard’ for all but atiny fraction of the people involved.

      1. Twain Twain

        Let’s give a real-life example of a hard problem in technology that prevents most entrepreneurs from tackling it unless:(1.) They have the tools and resources, particularly the resources of intelligence.(2.) They have either invented / custom-built the tools and /or have access to the tools via licensing them and so can deploy the tools in solving the hard problem.A hard problem would be the one that Google Brain, IBM Watson, Facebook’s new AI-Machine Learning are all investing $ billions in.How to program the machines to understand the meaning of what we type and the content we upload.”Hard to do” encompasses both the mathematical and technical knowhow as well as the resource investment available.It is comparatively easy to build an app that might result in a startup being valued at $ low millions. It is HARD to build infrastructure for human-like intelligence in the machines.

        1. Emily Merkle

          I find intriguing this common thread that a start-up must be a “build”. In my little corner of startup land, it’s about knowledge and innovation. Thinking and executing differently than everyone else sets a srartup apart. Yes, there is a tech infrastructure element, but it is not what brings in revenue. People do. My startup world is not populated by VC money. And that’s fine by me. Bootstrapping through client relationships builds character. But that’s just my reality; it’s all I know.

          1. Twain Twain

            “Build” can be the technical platform and/or client relationships, R&D, sales pipeline, educational & marketing materials and other processes and people skills which turn an idea into revenue.A film production is a startup too. As is a collective of people who decide to knit customised products for others and get those products marketed, distributed and sold.The reason startup has been associated with “build” is because of the Product Mgmt frameworks wrt prototypes, MVPs, scalability.It’s entirely feasible to be classified as a startup by combining knowledge and innovation in the form of sector-relevant content distributed via existing tools like WordPress, YouTube or Open Source libraries.It’s not the case that the lack of coding, database architecting and stack “building” means the founder isn’t doing a startup.A startup in the broadest sense of the word is simply having the optimism to begin on an adventure from idea to rewards (could be revenues or could be recognition / altruism / educational etc.)

        2. sigmaalgebra

          > A hard problem would be the one that Google Brain, IBM Watson, Facebook’s new AI-Machine Learning are all investing $ billions in.Actually, we are not in literal disagreement.Are there some hard problems? Sure. Hard for everyone? Sure.What I responded to was your> If nobody’s building it yet…It’s probably really HARD to do.and my response was:> Maybe not!> How ‘hard’ something is to do can depend a lot on what tools, etc, are brought to the challenge. Something that is really ‘hard’ for nearly everyone might be quite easy for someone bringing better tools to the challenge.Here’s where we likely differ:An old lesson in success in solving problems is good ‘problem selection’.E.g., for your> A hard problem would be the one that Google Brain, IBM Watson, Facebook’s new AI-Machine Learning are all investing $ billions in.Well, my view is that for a long time and likely still ‘artificial intelligence’ (AI) that is at all significantly ‘intelligent’ is poor problem selection. I don’t believe that anyone, especially at Google, IBM, or Facebook, has even as much as a weak little hollow hint of a tiny clue about how to do that. For such companies to spend “billions” is fairly easy; knowing how to do it effectively is much more difficult.”We can dance ’round and ’round and suppose, while the secret sits in the middle, and knows.”.Part of the value of a good ugrad degree in pure mathematics is the nearly endless series of lessons in the fact that there can be problems with very clean and solid solutions that, however, are just maddeningly difficult to find. So we learn that in some cases there really are very clean and solid solutions but that very likely we are not clever enough to find them. Thus it’s easy for me to believe that no one at Google, IBM, or Facebook is nearly clever enough to find the solutions to AI.IBM? Once I was at the IBM Watson research lab in Yorktown Heights and did some research. IBM did their best to review the research and concluded that it was “not publishable”. Suspicions confirmed: Since there was an error in the paper, essentially just typographical, just in an example, and not important to the paper, and IBM did not find the error, basically IBM was not able to find anyone who could read the paper at even an elementary level. IBM’s efforts to read my paper were totally incompetent. Indeed, the paper published routinely, with only minor editing, e.g., indenting the first line of paragraphs, in the first journal to which submission had been made. It remains a nice paper and likely useful. As I intended, the paper solves a pressing practical problem in large server farms and networks. At Morgan Stanley, the main Unix system administrator exclaimed, “We can use this right away!”. IBM blew it. My respect for the technical competence of IBM is way down below sea level. That IBM could do something solid in AI strikes me as just absurd.And I have little or no more respect for the chances of Google or Facebook.Basically nearly all the people working on AI are from ‘computer science’, and I have low to no respect for that field. Why? Their basic methodology is junk. In simple terms, nearly everyone in computer science is missing the most important tools for their work, a good ugrad and MS education in pure and applied mathematics. Right: If interested in research in computer science, essentially f’get about much in current computer science and concentrate, instead, on preparation, especially in pure mathematics. Why? Because mathematics has rock solid accomplishments, powerful tools, and the most powerful methodology in all of civilization. Sorry ’bout that. I know: Nearly every ugrad student concludes that it’s easier to major in English, economics, sociology, psychology, or basket weaving. Sorry ’bout that!There is just essentially nothing in computer science that is at all promising as a basis for progress in AI. Sorry ’bout that. Right: In AI software we may want to use AVL, red-black, or B trees. And if we have to sort some data, then we have some really good means. Terrific! Just one more $10 million DARPA grant and a few bright grad students and we’ll have it, right?So, again, that Google, IBM, and Facebook are spending “billions” to me is just next to meaningless for AI. They might get some publicity and turn out some software that gets used for something, but their chances of catching up with the intelligence of even a four week old brown tabby kitty cat are slim to none for decades.E.g., the group I was in at IBM was in AI, and we published a stack of papers. But my paper was the only one published in an archival journal not run by IBM. And, sorry, but my work was the only work of the group with any significant originality, power, or utility. My work: Original, correct, powerful, useful? Yes. AI? No!F’get about AI; unless clearly have some good ideas, then it’s poor “problem selection”. I’ve seen a lot of really bad ideas pursued seriously as AI; I have yet to see anything in AI even as good as that paper I published. So far AI is a junk field.Once our group gave a paper at an AAAI IAAI conference at Stanford. It was clear: The AI work was junk. The good work was just good cases of traditional engineering.So, maybe DARPA will be able to get self-driving trucks that can move supplies from Basra to Baghdad across exploding IEDs without harm to US soldiers. Good. Good engineering? Yes. AI? No.Can research make progress, including in practice? Yes. Example: The research for penicillin, streptomycin, and terramycin that cured my tummy ache. There was some more good research that cured my chronic middle ear infections. And there have been a few additional cases.My current project should be another example. Original, powerful, useful, valuable? That’s my intention and belief. Do well handling and approximating some of human ‘meaning’? Yes. AI? No. Indeed for a solution to my problem, I didn’t bother looking at AI. I mean, why just totally waste time? AI, so far likely and apparently there’s just nothing there yet. Some things DARPA throws money at work great, but not everything does.> How to program the machines to understand the meaning of what we type and the content we upload.Yes, software to “understand” ‘meaning’ is a holy grail problem in computer science, AI, or just any field that wants to attack the problem. And my work makes some nice progress on ‘meaning’, but for that progress I have a much more focused goal and, net, a solvable problem. For any significant progress on the problem you stated, I see no real hope for decades. Did I mention the importance of good “problem selection”?Here are two lessons: First, early in Rudin, ‘Principles of Mathematical Analysis’ is an exercise: In a topological space, a ‘perfect’ set is a set that is closed and every point of the set is a limit point of the set. Show that in a separable metric space, every closed set is the union of a perfect set and a set that is at most countable. Second, early in Rudin, ‘Real and Complex Analysis’ is the exercise, show that there are no countably infinite sigma algebras.Working with these two exercises can ‘build character’! The first one I struggled with in ugrad school for about 18 hours a day, neglecting everything else, for about two weeks. Finally a prof took pity on me and gave me a hint, consider points each neighborhood of which contains uncountably infinitely many points of the set. Presto: I got the solution in about 90 seconds! The exercise is now notorious and easy to find on Google. So, for the second exercise, in grad school, I was smart enough to consider uncountability, which is the key, and was the only student in the class who got the exercise.More ‘character building’: For positive integer n and the set of real numbers R, consider finite dimensional real Euclidean space R^n. Then the intersection of finitely many closed half spaces is a ‘polytope’ is convex and has flat sides and sharp ‘extreme’ points joined by straight edges, much as in R^3. Now consider the cone where all the components are greater than or equal to zero. For that cone, there is a polytope that, used as many identical ’tiles’, can ’tile’ the cone. The number of flat sides on this polytope is essentially an exponential in the dimension n of the vector space — really big bummer! Extra credit if can really visualize one of these tiles for n = 100,000 and with roughly e^n faces!From that tiling can appear to be an easy solution to the NP-complete problem of a polynomial algorithm for the traveling salesman problem. Yup, we’re diving into wildly astounding complexity. Sorry, if there’s a solution in there, then it’s deeper than I saw! So, I didn’t get the $1 million prize of Clay Math in Boston. The question of P versus NP is intricate and difficult beyond belief — real ‘character building’.So, I’ve seen in some clear terms how difficult some clearly stated problems can be. Such problems have done some ‘character building’ for me. E.g., I know too well that there are some just astoundingly difficult problems out there, even ones with really clear, short, easy statements. So, I’ve been warned: Good problem selection is just crucial, and bad problem selection, if not set aside, can lead to a lifetime of failure. So, be careful to avoid bad problem selection.I regard AI as just really bad problem selection. With some really good ideas, maybe AI will become good problem selection, but I’ve seen too often how difficult it can be to have such good ideas: Possible? Yes. Difficult? Commonly difficult beyond belief. Warning: In research into the unknown, there are monsters out there. So, don’t pick a problem just because a good solution would be nice to have (even if Google, IBM, and Facebook are spending billions on it). Instead, one part of good problem selection, at least before investing any significant resources, is to have at least some well founded hope that can get a good solution.AI: Nice to have a good solution? Likely! Well founded hope of getting a good solution? Nope! Conclusion? F’get about AI at least until there is some hope. Want to play with the AI problem instead of, say, watching the boob tube? Fine. Have fun. Odds are better than Vegas! Invest significant resources before good hope? Nope.

          1. Twain Twain

            Haha, well it may be a good thing I am a…MATHS graduate and I don’t buy into the good “problem selection” practices of various AI giants.Re. countability, most of the social media metrics falls into this category (number of “likes”, shares, node connections/edges etc).Uncountability gets us into the terrain of Godel and Quantum Information with Hamiltonian operators and is where it’s HARD but really really interesting.Mathematics does indeed have “rock solid accomplishments, powerful tools, and the most powerful methodology in all of civilization”.However, it is precisely the fact that it is a language for counting and quantification rather than a language like others for comprehension and qualification that………….The current AI predicated on the mathematics of logic, group set classifications and probability & stats cannot be “intelligent” like we are.That’s because our natural intelligence extends well beyond logic, rationality and probabilistic risk mgmt.

          2. sigmaalgebra

            A set is ‘finite’ if for some integer n the set can be put into 1-1 correspondence with the set {1, 2, …, n}.A set is ‘infinite’ if it can be put into 1-1 correspondence with a proper subset of itself.A set is ‘countable’ if it can be put into 1-1 correspondence with the integers or a proper subset of the integers.A set is ‘countably infinite’ if it can be put into 1-1 correspondence with the integers.A set is uncountable (uncountably infinite) if it is infinite and cannot be put into 1-1 correspondence with the integers.The integers, rationals, and algebraics are all countably infinite.The union of a countable set of countable sets is countable (proof essentially the Cantor diagonal process).Given a set A, let the set P(A), the ‘power set of A, be the set of all subsets of A. Then A and P(A) cannot be put into 1-1 correspondence. So, easily can generate a countably infinite sequence of sets no two of which can be put into 1-1 correspondence. So, roughly there are at least countably infinitely many infinities.The set of real numbers is uncountably infinite (the proof is another application of the Cantor diagonal process).A big question was, can the real numbers be put into 1-1 correspondence with the power set of the integers? The answer: In terms of Zermelo-Fraenkel set theory, the answer is undecidable. See work of K. Goedel and P. Cohen.

          3. Twain Twain

            The bigger question is what if the set contains entities that cannot (have not yet been) enumerated?These entities not being reducible to binary 0,1 or Yes/No.

          4. sigmaalgebra

            > The bigger question is what if the set contains entities that cannot (have not yet been) enumerated?> These entities not being reducible to binary 0,1 or Yes/No.In set theory, a ‘set’ consists of ‘elements’. We are supposed to know already what the elements are. And, to get around the Russell paradox, we have to have the elements in mind and fixed before we start to define sets.So, if we are unsure what the elements are, then we are not considering set theory.We agree that given a set A and an element x, either element x is in set A or it is not. That is, in your terms, the statement “x is in A” is either true or false, 1 or 0. It may take us some effort to determine if “x is in A” is true. If we are careful, then maybe in some case we can conclude that the statement is undecidable, but that is very tricky stuff and rarely encountered.In general in set theory, being “enumerated” plays little or no role. Informally, ‘enumeration’ is ‘counting’ as in 1, 2, 3, …, that is, stepping through the set of positive integers, that is, the set of ‘natural’ numbers {1, 2, 3, …}.A set that is uncountable cannot be ‘enumerated’: Would have to step through all the natural numbers and still would not have hardly even begun to ‘enumerate’ all of the set.Mathematics is just awash in crucially important uncountable sets, the most important of which is the set of real numbers, and, then, of course these sets cannot be ‘enumerated’.G. Cantor struggled with these issues in about the 1880s or so. Calculus, Fourier theory, Hilbert space, Banach space, probability theory, etc. are all much easier to swallow and digest than tricky questions about set theory down in the cold, dark sub, sub basement of mathematics, really, down in the wet dirt with the black beetles!The Russell paradox was from, as I recall, the early 1900s. Set theory then struggled for about 40 years before the relatively good clean up of the now broadly accepted Zermelo-Fraenkel formulation. An excellent, elegant, introduction is P. Halmos, ‘Naive Set Theory’. A more careful treatment of all the details is in P. Suppes, ‘Axiomatic Set Theory’. A brief, but nearly unreadably succinct, treatment of axiomatic set theory is in an appendix of J. Kelley, ‘General Topology’.Borrowing from the movie ‘Lawrence of Arabia’, Kelley, also as in his ‘Topological Vector Spaces’, “has a hunger for desolate places’.The ‘continuum hypothesis’ asks if the power set of the integers can be put into 1-1 correspondence with the reals, and K. Goedel struggled with this question, as I recall, in the 1930s-40s or so. P. Cohen did his contribution near 1964. Since then the subject has become ‘model theory’.One way to totally blow your mind is to get into the results that follow from accepting the axiom of choice, i.e., Zorn’s lemma and maybe another dozen equivalent statements. One consequence is that there is a subset of the real line that is not Lebesgue measurable — just astounding. The construction takes some cleverness.BTW, I still have Max Zorn’s copy of Paul Cohen’s paper!For another example, in general topology, the Tychonoff product theorem states that the product of compact spaces is compact, and this result is equivalent to the axiom of choice.I spent too much time in the cold, dark sub sub basement of mathematics. Apparently considering these questions didn’t do much good for Goedel — I suggest staying out of that part of mathematics.

  19. sigmaalgebra

    So, let’s see:(1) It’s hard being a CEO because may have afamily and, then, have to neglect family becauseof the stress when have to fire people.(2) It’s hard being a CEO because have to makehard decisions, that is, all of cases A, B, and Care close and bad.(3) It’s hard being a CEO because the business mightfail and might lose job.Hmm ….For (1) the employee that gets fired might have a family; the CEO still has a job; the formeremployee doesn’t; believe me, in a firing, what’s hard is on the employee.For (2), if the cases A, B, and C are not equallybad, then pick the least bad. Else the three arenearly equally bad and it’s easy to make nearlythe best possible decision, just roll a die and assign 1-2 to A, 3-4 to B, and 5-6 to C. Hard?I don’t think so!For (3) before the business fails and the CEO loseshis job, no doubt most of the employees lost theirjobs long before. So, the CEO was the last to goand likely had the best salary, golden parachute,etc. What’s hard is on the employees, not the CEO.Some of these thoughts were from a review of thebook at Amazon.Anzio, Salerno, Omaha, Kasserine were ‘hard’?The A-bomb was ‘easy’ How do we know?Because Anzio, Salerno, etc. were disastersor nearly so and, thus, ‘dramatic’. The A-bombworked essentially just as planned, with relativelylittle in blood, sweat, tears, ‘hard’ decisions, failures that nearly brought disasters, improbableextractions of victory from the jaws of defeat,and, from the work at Oak Ridge and Hanfordto the first test and the two deployments, everythingworked as planned — not much drama exceptfor the final results themselves.I remember the NBC broadcast of the Summer Olympic basketball game with thefirst Dream Team — Magic, Larry, etc. –playing the Chinese team where the NBCannouncer said, “The outcome here isnot in doubt. We will cut away now butreturn in case of interest.” or some such.BS!!!!! I quit watching, got the FAX numberof the head of GE over NBC, and sent anangry letter.Of course the game was not in doubt. That’snot a bad thing but a very good thing! A closegame means that neither team did a reallyexceptional job and is a bummer! Instead, theDream Team did a fantastic job and was a joyto watch. Indeed, got to see some things usuallydon’t see: A Chinese guy was bringing theball up court; two Dream Team guys signaledeach other; one of the DT guys stole theball at half court and did an overhand passto the second DT guy who in a few stepsdid a grand slam dunk! Piece of cake! Easy2! The Chinese coach had a huge smile atthe lesson he and his team got in DTcompetence!Sorry, I don’t like the hanging by the fingernailsdrama and, instead, want the overwhelming,result not in doubt, DT or A-bomb. Sorry,Ben. Sorry about Opsware. Good thatyou pulled victory from the jaws of defeat,but you remind me more of Anzio than Los Alamos, more of Mark Clark thanRobert Oppenheimer.

  20. Joe Wallin

    I’m listening to it on audio. Great book.

  21. Simone Brunozzi

    Fred, you might also want to take a look at my review of the book, “The Ugly Thing about The Hard Thing about Hard Things.”

  22. Harry DeMott

    You may have seen Ben in action, but have ou seen him breaking it down with Nas at the A Grade party at SXSW. Classic stuff.

  23. Sam Parker

    I’m only partway in after picking it up this weekend due to a good machine recommendation on Amazon.One bit that stuck out to me was his commentary about calculus vs. statistics in a Peter Thiel reference (… etc), picking calculus over Thiel’s statistics worldview.I appreciate that coming from a place where we often have to just make something work and the determinate calculus approach is about getting from A to B, sometimes just by force of will. I sometimes have to remind myself since my analytics background gets me into the statistics mindset.

    1. Joe Wallin

      I just finished that part. I like the writing about what the struggle is. Good writing.

  24. Manuel Molina

    i’m not a VC, but learning something new everyday without any apparent practical benefit is a great habit.just bought the book, thanks for sharing.