Wil is a great writer and worth reading. In this case, I'm going to raise objections to his thesis, but I have tremendous respect for his credentials. That said, I think the comparison to mining is a stretch, also farming, but lets start with mining.
Wil makes it sound like most mines blow up or explode at the first touch of a nugget of whatever you're trying to uncover. Or alternately that after taking investor money and building all the mine infrastructure, you sell it off as soon as you strike it rich. I would think the opposite is true, the last thing the mine owner would want to do is flog it off just when it starts making big money.
Eventually mines run out this is true, but sometimes they last for decades if not hundreds of years (the hundreds of years was probably more common before modern methods).
Secondly, with the mines if you read Jared Diamond's† book, I think it is called Collapse, he talks about this in great detail. And the biggest problem with mining is actually what happens when the mine closes. According to his research, nowadays the mining company is required to estimate the cleanup costs, and over the lifetime of the mine they contribute to a fund to cover the cleanup cost. Sounds good, right? The problem of course is that it is the company that estimates the costs, and they have an incentive to grossly underestimate the costs of cleanup. The usual discrepancy is something like two orders of magnitude. Ie assuming a large mine the company estimates costs of 10 million, hands the state govt a cheque for that as it closes shop. Then when the state govt discovers the actual cost is approx 1 Billion (maybe more) the company (which has long paid out all its funds as dividends to its shareholders) shrugs and declares bankruptcy.
I don't think that there is really any parallel in IT in terms of cleanup costs. You could argue that Y2K was similar, but I would point out that wasn't a uniquely mining only problem. All kinds of software was effected, from mining (startups on steroids), farming (slow and steady startups, the tortoise vs the hare), vendors, universities, consultants and even your big software retailers (Microsoft) all had this problem.
Okay, he admits he's stretching the metaphor (on the rack as it were). :D
What about farming? Personally I'm a fan of farming in the software sense, but farming in the real world is a nightmare. Assuming you don't manage to destroy the land, farmers I've talked to say that in good years they are making ~4% (return on capital), and in bad years they are losing 2-3%, and there are more bad years than good years. How do they survive? Typically by taking loans out against the (ever increasing) value of the property. But if the value of the property doesn't keep going up, or if the banks decide not to loan the farmers money, or if there is a particularly bad drought... then farming is much worse than mining and you get farmers walking away from their land, or just dying on it. Jared talks a little bit about how bad mining is, but the main thrust of his book is that improperly managed farming KILLS ENTIRE CIVILIZATIONS. In terms of cost/benefit, I think dirty water from mining >>> extinction. :D
He touches on another metaphor, the Lottery. Lotteries are usually regarded as a bad thing (a tax on people who are bad at math). The problem is that for every $1.00 you put in you get back out say $0.40 on average. In other words, all lotteries have a negative expectancy (with the notable exception of the UK Lottery, because even though you are unlikely to win, at the end of the day you can still get back all the money you spent on it (it is more similar to a savings account with pathetically low interest, and a miniscule chance of some excitement).
Okay, fine, lets take it as given that 'lotteries' with negative expectancy are bad....
If the 'software mining' (ie startup) industry was a lottery, would it have a negative expectancy? I suggest probably not. What if there was a lottery that for every $1 ticket there is a 1% chance of winning $200? That'd be a positive expectancy! I'd play that, you'd have to be bad at maths not to. Moreover, even if it was break even (ie for every millionaire ten people have to gamble and lose their house), as the mining example shows, sometimes there are hidden costs/benefits. I'm guessing that the startup 'lottery' has a positive expectancy (especially in silicon Valley, perhaps not everywhere else in the world), but if someone argued that the benefit to society of all the startups is negative I'd think they were crazy.
An example: I've heard it said that since the beginning of commercial flight in the US that the industry as a whole has barely (or not even) broken even. But to suggest that I personally don't benefit from the ability to fly round the world in 30 hours as compared to sailing round the world in six months is ludicrous. Even if the airline industry didn't make their average investor rich, it provided more than enough benefit to society to be able to justify its existence.
Lastly, and I suspect this is very wall of text so I apologise, there is a logical flaw underpinning Wil's entire thesis. He assumes that somehow the IT industry is special. He assumes if you launch a software company with an exit strategy that is somehow different (bad) compared to say launching a furniture company with an exit strategy.
I think that rather than just assuming that software is different in this regard, I think he should at least try to make an argument to that effect.
†If the name is familiar, he wrote Guns Germs and Steel as well, which is one of those books that lots of people have heard of, but few have read :D
Wil makes it sound like most mines blow up or explode at the first touch of a nugget of whatever you're trying to uncover. Or alternately that after taking investor money and building all the mine infrastructure, you sell it off as soon as you strike it rich. I would think the opposite is true, the last thing the mine owner would want to do is flog it off just when it starts making big money.
Eventually mines run out this is true, but sometimes they last for decades if not hundreds of years (the hundreds of years was probably more common before modern methods).
Secondly, with the mines if you read Jared Diamond's† book, I think it is called Collapse, he talks about this in great detail. And the biggest problem with mining is actually what happens when the mine closes. According to his research, nowadays the mining company is required to estimate the cleanup costs, and over the lifetime of the mine they contribute to a fund to cover the cleanup cost. Sounds good, right? The problem of course is that it is the company that estimates the costs, and they have an incentive to grossly underestimate the costs of cleanup. The usual discrepancy is something like two orders of magnitude. Ie assuming a large mine the company estimates costs of 10 million, hands the state govt a cheque for that as it closes shop. Then when the state govt discovers the actual cost is approx 1 Billion (maybe more) the company (which has long paid out all its funds as dividends to its shareholders) shrugs and declares bankruptcy.
I don't think that there is really any parallel in IT in terms of cleanup costs. You could argue that Y2K was similar, but I would point out that wasn't a uniquely mining only problem. All kinds of software was effected, from mining (startups on steroids), farming (slow and steady startups, the tortoise vs the hare), vendors, universities, consultants and even your big software retailers (Microsoft) all had this problem.
Okay, he admits he's stretching the metaphor (on the rack as it were). :D
What about farming? Personally I'm a fan of farming in the software sense, but farming in the real world is a nightmare. Assuming you don't manage to destroy the land, farmers I've talked to say that in good years they are making ~4% (return on capital), and in bad years they are losing 2-3%, and there are more bad years than good years. How do they survive? Typically by taking loans out against the (ever increasing) value of the property. But if the value of the property doesn't keep going up, or if the banks decide not to loan the farmers money, or if there is a particularly bad drought... then farming is much worse than mining and you get farmers walking away from their land, or just dying on it. Jared talks a little bit about how bad mining is, but the main thrust of his book is that improperly managed farming KILLS ENTIRE CIVILIZATIONS. In terms of cost/benefit, I think dirty water from mining >>> extinction. :D
He touches on another metaphor, the Lottery. Lotteries are usually regarded as a bad thing (a tax on people who are bad at math). The problem is that for every $1.00 you put in you get back out say $0.40 on average. In other words, all lotteries have a negative expectancy (with the notable exception of the UK Lottery, because even though you are unlikely to win, at the end of the day you can still get back all the money you spent on it (it is more similar to a savings account with pathetically low interest, and a miniscule chance of some excitement).
Okay, fine, lets take it as given that 'lotteries' with negative expectancy are bad....
If the 'software mining' (ie startup) industry was a lottery, would it have a negative expectancy? I suggest probably not. What if there was a lottery that for every $1 ticket there is a 1% chance of winning $200? That'd be a positive expectancy! I'd play that, you'd have to be bad at maths not to. Moreover, even if it was break even (ie for every millionaire ten people have to gamble and lose their house), as the mining example shows, sometimes there are hidden costs/benefits. I'm guessing that the startup 'lottery' has a positive expectancy (especially in silicon Valley, perhaps not everywhere else in the world), but if someone argued that the benefit to society of all the startups is negative I'd think they were crazy.
An example: I've heard it said that since the beginning of commercial flight in the US that the industry as a whole has barely (or not even) broken even. But to suggest that I personally don't benefit from the ability to fly round the world in 30 hours as compared to sailing round the world in six months is ludicrous. Even if the airline industry didn't make their average investor rich, it provided more than enough benefit to society to be able to justify its existence.
Lastly, and I suspect this is very wall of text so I apologise, there is a logical flaw underpinning Wil's entire thesis. He assumes that somehow the IT industry is special. He assumes if you launch a software company with an exit strategy that is somehow different (bad) compared to say launching a furniture company with an exit strategy.
I think that rather than just assuming that software is different in this regard, I think he should at least try to make an argument to that effect.
†If the name is familiar, he wrote Guns Germs and Steel as well, which is one of those books that lots of people have heard of, but few have read :D