Easily, if it gets larger because people are naturally following other incentives. How does the Pacific garbage patch keep getting larger if people are not focused on intentionally making it larger?
> since they usually have largely the same effects.
They absolutely don't. I'd you have "not enough money" to meet your current monthly budget that means you can't afford to put anything aside for your vacation next year but can still get by.
If you have "no money", you can eat or pay your rent.
The two phrases have entirely different meanings, and usually with important differences in connotation.
You don't have to understand how an economy works to participate in it. It helps, for sure, but it isn't required. Trial and error goes a long way. A lot of companies will go out of business because they are doing something that doesn't work, but some will chance upon something that works.
I think what Gates is saying is that, while the "bits and bytes" economy is getting larger, perhaps lawmakers are still thinking in the terms of the "old model" which Gates indicates is obsolete.
He then says:
That has major implications for everything from tax law to economic policy to which cities thrive and which cities fall behind, but in general, the rules that govern the economy haven’t kept up. This is one of the biggest trends in the global economy that isn’t getting enough attention.
It could simply be either from the study he mentioned or be worded as such to make it easier for non technical people to understand, as in more are familiar with MP3 than AAC. It is irrelevant to the greater point of the article
I'm sure a man as smart as Bill Gates could find a way to write it in a way that is accessible to non-technical people and still not so egregiously factually incorrect.
I think his straw man for supply and demand is off. Higher demand doesn’t create lower prices. Lower prices creates “more quantity demanded” at a given level. If demand increases, then prices and quantity both go up. This may be semantics, but it isn’t like economics completely falls apart. Similarly, economics is able to handle products with high fixed and low marginal costs. (Natural monopolies)
I suspect he gave this book review assignment to an intern that majored in something besides CS, business or Econ.
>> ... as demand for a product goes up, supply increases, and price goes down
I think it is fine but confusing because it skips over some steps. A longer version might be:
... as demand for a product goes up, the demanded quantity goes up, which causes supply to increase (to meet the higher demanded quantity) which then pulls the price down (due to the increased size of the market and competition).
The net effect of this is decreased price overall due to economies of scale (since the total production volume of the product in the market is now higher).
That's bothered me before. It really seems like economists flipped their axes on a basic supply demand curve.
They discuss it as if price is the independent variable and quantity (demanded or supplied) is the dependent, and yet any 9th grader would be marked down for putting their independent variable on the "y-axis".
It amazes me how something generally accepted as wrong stays that way.
I had the same thought.
Also, I'm not even sure that this diagram makes sense overall because supply, demand and price are so tightly interconnected (with feedback loops in both directions) that it doesn't make sense to separate them like this.
> Also, I'm not even sure that this diagram makes sense overall because supply, demand and price are so tightly interconnected (with feedback loops in both directions) that it doesn't make sense to separate them like this.
How so? Econ major here, and I agree the axes should be swapped, but I don't follow your other point.
Why is it that when the price drops, consumers would want to buy more and producers would necessarily want to produce less?
If I was a company director and the price of the commodity that I produced dropped by a lot, I would work harder to produce more of the commodity to offset the lower profit margins; that way I could earn the same bonus at the end of the year in order to make the payment for the mortgage on my yacht.
Also in the case of cryptocurrencies, it doesn't seem to work the way the graph suggests; when the price goes up, people buy more.
Ah, I think you have the same misconception as Gates in the post, of confusing the concepts of "points along the curve" with the curve itself.
> Why is it that when the price drops, consumers would want to buy more and producers would necessarily want to produce less?
I would phrase it as when the price is lower, there are more consumers who would buy, and fewer producers to sell. To take the demand curve, for example, it doesn't explain one person's behavior at different prices. Rather, think of the curve as a thousand points, each representing a different person and their own maximum price at which they would buy the item. The supply curve is a thousand producers each with the minimum price they'd be willing to operate at.
> If I was a company director and the price of the commodity that I produced dropped by a lot, I would work harder to produce more of the commodity to offset the lower profit margin
Yes, this is already reflected in the supply curve. If your company is able and willing to produce that commodity at a lower price, you're already reflected on the supply curve as one of the points on the line below the current equilibrium price. You have what's called a "producer surplus" if you'd be willing to produce at a lower price than the equilibrium.
Take the supply of uber drivers for example. If the price of rides decreases, you may to some extent see drivers adjusting how much they drive, but the main effect is to remove drivers from the market who have better uses of their time, less efficient cars, etc. Similarly, if the price increases you'll see new uber drivers coming online driving their clunky SUVs or whatever.
Or, in the case of cryptocurrency, the supply follows this as well: at a higher price, less efficient miners or people in locations with higher utility prices will be able to profitably mine, and so the number of people supplying computation will increase.
> Also in the case of cryptocurrencies, it doesn't seem to work the way the graph suggests; when the price goes up, people buy more.
This, too, works with the supply and demand curves, when you understand them properly. The concept at work, here, is that changing supply or demand (i.e. think of the thousands of people that make up each curve shifting their preferences in the aggregate) represents shifts of the curve, not movements along it. Increased demand is the curve shifting up and/or to the right (it's the same thing).
So, when the price goes up, people see it in the news and have FOMO and so "demand increases", meaning the curve shifts to the right. If you look at the graph again, you'll see that means the equilibrium price increases, which is what's happening. That, in turn, drives more breathless speculation and aggregate demand increases again, pushing the curve further to the right, driving up the price some more.
Concept check: the supply curve is not shifting, here. Per my earlier paragraph, the increasing price may mean more miners come online, but that's already what the supply curve means. A shift in the supply curve would be something like a new breakthrough energy technology making everyone's utilities cheaper. That means if everyone's minimum price before was $x, they're now willing to supply computation at $x + 5, shifting the supply curve to the right and (if you check the diagram) driving down the equilibrium price (since the # of bitcoins in circulation will increase faster and the demanded fee by the miner's will be less).
It's quite common to have a situation where in the short term y is a function of x, and in the longer term equilibrium-x is a function of ambient-y.
The cylinder-with-a-piston used to teach Boyle's law in secondary-school physics is a familiar example: in the short term pressure is a function of volume; in the longer term volume is a function of (external) pressure.
I would say there's no independent variable here because the effect goes both ways. Suppliers set their prices based on the quantity demanded, and consumers choose a quantity to purchase based on the price set.
> That's bothered me before. It really seems like economists flipped their axes on a basic supply demand curve.
Given that any Y-X chart is an X-Y chart if rotated 90 degrees and then flipped horizontally, is there any real issue here, other than the axes being the reverse of the standard statistical style?
There is a lot wrong with the supply and demand description including lack of elasticity and scalability as a factor.
Software is the extreme end of scalability. One good lesser example is manufacturing vs repair. As mass production took off in the time it took to repair one item a worker could make ten new. Wasting some material on defectives made more sense than repairing them usually unless the cost of failure is extra high, inspections more expensive or a more cautious approach yielded better viable output. It is only when the expense of new is high that repair remains intrinsically viable (if you enjoy repairing things economics don't matter as it is its own reward).
If fixed per unit costs are high extra supply does little good - especially if it is durable. You won't be able to buy $100 for $99.
“It’s a sunk cost. If your investment doesn’t pan out, you don’t have physical assets like machinery that you can sell off to recoup some of your money”
That also means it’s cheaper to develop. Software is so popular since it’s only requires a developer and a PC. The return/investment is a lot higher and faster than traditional hardware development. Cheap and fast is what everyone wants.
Of course investors are now realizing, in the case of self driving cars (or cars themselves with Tesla), that hardware still plays a role, and it’s not cheap and fast with instant riches.
20 years ago, biotech was supposed to be the hot thing after the internet. Instead it has been software services, focusing on service labor. Biotech is not cheap and fast enough.
Also, his statement implies that digital assets have no real-world financial value. In practice, even if the investment fails, the codebase of the company, body of active users, and any user data gathered will have some value.
This book review has a few economic fallacies and was probably not written by Gates. But, these points are interesting--
"It" refers to "intangible investment" here--
1. It’s a sunk cost. If your investment doesn’t pan out, you don’t have physical assets like machinery that you can sell off to recoup some of your money.
2. It tends to create spillovers that can be taken advantage of by rival companies. Uber’s biggest strength is its network of drivers, but it’s not uncommon to meet an Uber driver who also picks up rides for Lyft.
3. It’s more scalable than a physical asset. After the initial expense of the first unit, products can be replicated ad infinitum for next to nothing.
4. It’s more likely to have valuable synergies with other intangible assets. Haskel and Westlake use the iPod as an example: it combined Apple’s MP3 protocol, miniaturized hard disk design, design skills, and licensing agreements with record labels.
"This book review has a few economic fallacies and was probably not written by Gates."
Dunno if they're fallacies so much as oversimplifications. But the oversimplifications aren't really the point; they're just meant to reference the concepts and give a hyper-quick overview if you've never heard of them. The rest of the review remains a valid point if you substitute more realistic economic concepts. This is my pre-emptive reply to the inevitable dozens of posts arguing about the oversimplifications. None of it matters to the point the author wanted to make.
Note I say "valid", not correct. Whether economists are undervaluing intangible assets is a rich and interesting question that I have only vague opinions about personally. I'm just pointing out that the argument itself does not depend on the oversimplified economic concepts used to introduce the point.
> Dunno if they're fallacies so much as oversimplifications.
No, they're far enough off of correct that I wouldn't even call them oversimplifications. The statement of the problem is wrong:
The second assumption this chart makes is that the total cost of production increases as supply increases.
That's simply not what the curve says at all. The curve isn't about any one producer making more of an item (as the Ford example given), it's about how producers with different costs are able to profitably add to the supply or not, based on the price they can get for the product.
The review might be fine otherwise, but there's no reason to throw supply & demand under the bus to motivate it. In fact, if you frame it correctly (say, technical innovations have enabled more people to more easily write software, thereby shifting the supply curve to the right and lowering prices), you can still meaningfully analyze these situations with those curves.
1) The first two paragraphs about 'assumptions' are essentially false. There are no assumptions in supply and demand, it's not a dynamic equilibrium, it's a snapshot.
Yes - over time supplies tend to change due to demand, of course, but that's 'long run' stuff ... not in that chart. The author is kind of misinterpreting the chart.
2) It's not 'sunk cost' that's the issue, this is about 'fixed' vs. 'unit' cost. Software is all 'fixed' (i.e. setting up the production line) whereas most material things are mostly about 'unit' costs (i.e. the cost to make an item)
3) Software sales are a tiny fraction of the economy because almost nobody is selling software. People are selling 'services', basically products that 'do stuff using software'.
The 'key ingredient' is usually some type of business or experiential knowledge, not some algorithm.
4) Software development doesn't finish at v1, it's ongoing.
Yes and there's more that is wrong in the article:
"as demand for a product goes up, supply increases, and price goes down"
No, when demand goes up, the price will go up as well unless supply increases by the same amount. The price will only go down when supply increases and demand doesn't keep up with it.
"the tools many countries use to measure intangible assets are behind the times, so they’re getting an incomplete picture of the economy. The U.S. didn’t include software in GDP calculations until 1999. Even today, GDP doesn’t count investment in things like market research, branding, and training—intangible assets that companies are spending huge amounts of money on."
First of all let's skip the part about what the US did until 1999 because we're living in 2018 at the moment. And then the fact that market research, branding etc. is not included in GDP calculation has little to do with the digital age. There has always been market research and branding, not just now. One might argue that it should be included but that is really a separate topic.
"It took time for the investment world to embrace companies built on intangible assets. In the early days of Microsoft, I felt like I was explaining something completely foreign to people. Our business plan involved a different way of looking at assets than investors were used to. They couldn’t imagine what returns we would generate over the long term."
No doubt things were like that in the 80's. But just take a look at the most valuable Dow Jones funds now and it's clear that intangible assets are valuated well nowadays.
I think the issue is that the prose isn't clear about what 'an increase in demand' means. It can mean either a movement along the current demand curve due to a shift in the supply curve, or it can mean a shift in the demand curve, and those changes correspond to different things in the real economy.
"demand for a product goes up, supply increases, and price goes down"
Assumption 2 from article:
"The second assumption this chart makes is that the total cost of production increases as supply increases. Imagine Ford releasing a new model of car. The first car costs a bit more to create, because you have to spend money designing and testing it. But each vehicle after that requires a certain amount of materials and labor. The tenth car you build costs the same to make as the 1000th car"'
In assumption 1 - he's not describing how equilibrium works at all. Equilibrium is the price the market will clear at given a supply and demand curve. If 'demand goes up' - given a fixed supply demand curve ... 'price goes up'. So what the author must be referring to is some kind of market response, i.e. if demand is increasing for something, more suppliers will come along and build it, create more competition, and drive prices down. This has little to do with the Supply/Demand curve.
I don't even know what assumption 2 is about, I get what he is saying but it has nothing to do with supply and demand curves.
Anyhow - even if there is logic in his statements, neither assumption is drawn from the Supply and Demand curve he is showing.
He's referring to the static equilibrium in assumption 1. Supply & demand curves, as drawn in economic textbooks, refer to marginal supply and marginal demand. The X-axis is is quantity, so what the graph is saying is that producing 100,000 units of some good is assumed to cost more than producing 10,000 units, and similarly consumers will be more willing to buy 10,000 units than 1,000 units at a given price (the Law of Downward-Sloping Demand, for those who remember their intro microec).
With assumption 2, he's saying that in software (and many other businesses now), supply curves do not necessarily slope upwards. For software, producing 100,000 units takes exactly as much money as producing 1,000 units, and can be offered at the same cost.
The author (who is Bill Gates, BTW, and can be assumed to know a thing about microeconomics) only alludes to this and doesn't say it outright, but there are a number of knock-on consequences to flat supply curves that explain much of the counterintuitive properties of the tech industry but aren't well-explained by classical economics. For example, the "lottery economy" aspect (where there is one company that ends up dominating a market and the size of that company is dictated only by the size of the market) comes from zero marginal costs: once a software firm has built a solution that satisfies a market, there is nothing stopping it from expanding to serve the whole market far faster than any competitor can enter. The existence of the VC industry follows from this, because when success is binary and can't be interfered with once discovered, it makes sense to spread capital across a number of bets, hoping for a hit, rather than rationally analyze how much additional return invested capital will give in an existing business.
Google & Facebook's dominance also stems from this zero-cost-of-production property. When your cost of production to provide the good or service is zero, then the only thing blocking your continued expansion is finding & convincing potential customers. With CAC forming an increasing percentage of your total marginal costs, any services that lower it or make it scale better can realize huge profits.
> For example, the "lottery economy" aspect (where there is one company that ends up dominating a market and the size of that company is dictated only by the size of the market) comes from zero marginal costs: once a software firm has built a solution that satisfies a market, there is nothing stopping it from expanding to serve the whole market far faster than any competitor can enter.
I'm not sure what you mean, this is 100% covered in general economic theory with barriers to entry.
> Google & Facebook's dominance also stems from this zero-cost-of-production property. When your cost of production to provide the good or service is zero
But it's actually not 0 or even close to it. There are still economies of scale at play, if Facebook were to lose millions of users their costs would grow per user, which is what a marginal cost curve represents. Just because the curve is different doesn't mean it doesn't exist.
I'm pretty disappointed with Gate's post, because every academic economist I know is mocking it...since it's not well represented or thought out. It ignores so many principles and theories in economics for the sake of being "ground breaking." Even though most of the theories are hundreds of years old and still apply completely.
When I first learned that software had a price during my very young years, I was in quite disbelief. Isn’t it just bunch of invisible bits? Can’t it be copied with almost zero cost? Why someone would pay for it anyway when kids like me can make it? It’s hard to charge money for things people can’t physically posses. In early years, almost all software companies employed a trick to design big attractive boxes with bit of goodies. When Gates wrote BASIC, many programmers themselves in the community didn’t thought any software was worth paying for. It has taken almost a generation to internalize that bits have price and we have got rid of most of those boxes finally.
On the side note, crypto currency is the new new intengible assets that is hard to internalize. Ironically, Gates has wrote them off as not worth calling them assets because they aren’t store of a value.
There's a shrink-wrapped cardboard package that contains Windows 98. It's the size of a cereal box, and at least as colorful.
Today, I'd be intensely annoyed if I had to wait for and pay for a large box to ship - I'd much prefer an instant download, but the box was much more tangible than an email with a download link and license key.
At one point the packaging was required by the retailers, who wanted it to conform to a standard shelf size. Eventually it could shrink to CD or DVD jewel case size. Might be a discussion of this on filfre.net somewhere.
In the even earlier days, software fit on floppy disks. And it absolutely could be downloaded over a modem - it took about 1 hour to download 1MB over a 2400 baud modem, not much longer than going to the store.
The problem was that you couldn't convince consumers to pay for software downloaded over a modem. (The infrastructure wasn't in place anyway - nobody was going to send their credit card number unencrypted to a random BBS sysop, and encryption was outlawed as a munition.) Many software firms in fact had huge problems with piracy over modem, and would build copy-protection into their software so you needed the original physical disk to run the software.
The idea that you could pay for a license key that unlocks software you download over the Internet only happened in the early 2000s, driven by combination of remaining restrictions on encryption being dropped, easy-to-integrate payment solutions for webpages coming to market, and widespread broadband adoption meaning that most consumers could easily download a big file.
At the time of Windows 98 you probably wouldn't have wanted to download the damn thing.
If you are of a younger generation, you may not believe it, but many of us in those days made do with something called dial-up connections to the internet.
I well remember downloading the StarOffice - precursor to LibreOffice - suite in the days before it was taken over by Sun, meaning 1999 at the latest. Took a little over five hours, and was done at night when rates were lower. I actually stayed up, nursing the process along. And afterwards guarded the downloaded installation file, almost with my life, although it took up an inordinate amount of precious storage space. It was 50MB and a bit, you see.
Ah yes, the memories of those agonizingly slow (and noisy) Internet connections. Ironically, software, particularly games, still can take 5 hours to download. It's just that they're inordinately huge now. Battlefield 1 with its patches can run over 110 GB.
Games companies often had what they called "feelies": my copy of Ultima 7 came with a cloth map of Britannia and a small black "moonstone".
You didn't have this for business software, but you did need something to keep the disks in. I think the largest such thing I saw was 100+ floppy disks; here's Raymond Chen on floppy installs of Windows: https://blogs.msdn.microsoft.com/oldnewthing/20050819-10/?p=...
Well all games used to come as a CD in a box, simply because it wasn't viable to download them from the internet. I still remember the amazing box for GTA:SA that came with a large map and a little booklet to introduce you to the game.
The box designs were driven more by attracting attention from retail customers in physical stores than by perception of value. Just like cereal boxes.
The user manuals were absolutely necessary for complex software. Most users had only a single low resolution monitor back then so online documentation had poor UX.
> Imagine Ford releasing a new model of car. The first car costs a bit more to create, because you have to spend money designing and testing it. But each vehicle after that requires a certain amount of materials and labor. The tenth car you build costs the same to make as the 1000th car.
As someone with software and hardware at Ford in Dearborn and at many of their suppliers, this is not true. Ford knows that the costs decrease over time, and demands part of this cut from their suppliers. The contract includes a decreasing price over time or over quantity, because ostensibly you're increasing efficiency and writing off initial engineering costs.
I imagine that Microsft does the same with their hardware division, and agree with the sibling comments that Gates probably did not write all of this.
The fact that replicating a piece of software or music digitally has near zero cost suggests that the fair price according to classic economic theory is zero (the argument being, that the price in a competitive market should approach the limit price). Of course what we see instead is a lot of companies that are now rent seeking and that no longer sell a piece of software to you but rent it out for a fixed price (SaS, Adobe Creative Cloud, Jetbrains, season passes in games etc.), similarly with DRM and Music/Movies/eBooks Spotify/Netflix/Amazon. Uber also fits this pattern in that they rent out capital that other people provide to them for free (insane if you think about it).
What hopefully will happen over time is that these rent seeking entities are squeezed out of the market over time, otherwise we are entering an area of corporate feudalism. This can be prevented as long as the assets needed for production are easily and cheaply available, which is the case thanks to the Free Software Movement.
I think you are confusing the definition of rent seeking, meaning achieving economic gains by means of regulation that is virtually risk free, with collecting rent as a form of payment for a service or product. These are two different concepts, despite the name they share.
Photoshop, being nearly a monopolist, seems to almost fit the definition. Regulation is the only thing here that doesn't match, but being a de facto industry standard would be close enough for me.
"rent seeking", in economics, is not limited to regulation. We would distinguish between those who earn money by creating new value - as the inventor makes some process more efficient, and so makes his living - and those who earn a living despite creating no value. The easiest example of this is the man who has inherited some plot of land, and rents out the plot to someone. He might earn a healthy living - but where has he created value? The strongest example of this is perhaps he who earns money by destroying value - say the window repairman who goes out at night and smashes windows - but this is not required to be classed as rent seeking.
Software costing a few hundreds of dollars frequently provides tens of thousands of dollars of value.
This is quite similar to "free" (as in beer, let us not go down that tedious path). Say it only provides thousands of dollars of value and it's still pretty similar.
>that the fair price according to classic economic theory is zero
I don't think any economic theory is that naive.
It's actually the cost of production divided by the number of sales, where the cost of production can be many millions of dollars worth of development and testing and the size of the addressable market can in some cases be very few.
Adobe products are continually worked on. Most subscription software is. You get continued support and feature updates, and the subscription model makes more sense for that kind of development than the yearly waterfall version releases.
JetBrains has the best of both worlds with a fallback license to let you keep using an old version, or a subscription to get continued updates.
Games are becoming increasingly long lived pieces of software as well. With bugfixing and updates continuing for years. The industry is struggling to find novel ways to make money in that environment, which is why you get things like loot crates. I would prefer an honest subscription over that.
What I see is a devaluing of software as a product. I should be able to spend a couple of hundred hours on a package for a framework and sell it, but there is always someone offering a free version and then using weird and sometimes shady ways to make money. I wish it were more common to just sell code for money, instead of people pretending they are altruistic and then selling your data, selling your eyeballs or locking you into their ecosystem. Subscriptions are at least an open and honest transaction.
The price should approach the limit price if you assume there's no initial cost of making the software. And for most SaaS, there's still an upkeep cost the company must pay. Free software just isn't feasible in the real world past the things that can benefit everyone at a very low cost (where some entities will take it upon themselves to employ people to focus on, for example, the Linux kernel).
I generally agree with Gates on a lot of things, but there are some generalizations here that don't apply to a lot of software.
1. While the cost to produce N+1 and N+1000 version of the software is same in terms of producing the code, if you have a piece of software that functions more like a specialized tool, training user N+1 and N+1000 can be radically different and therefore more time consuming and costly.
2. Sunk cost of development: IP has a lot of value. If you work in producing custom solutions a lot of these solutions have components (code) that work across customer needs. Development may fail on one project, but helps in another. Overtime you have the components to sell a COTS
3. Goods actually aren't intangible, value is. Electricity costs money and it takes the infrastructure to support a small city to run a large-scale server farm. What's missing in the economic model are the costs we don't always factor in or more importantly that are publicly subsidized. Look at the breaks companies get at MSoft, Google, etc on utilities and land.
The laws have be setup so, all the value is trapped in the giant network companies. No liability for content. No portability requirements(you can port your phone number, but not your data or messaging addresses). Rent seeking is blocked for utilities, so they don't capture it from google, facebook etc..
The costs of producing N+1 and N+1000 should in fact be different. With N+1 you train the user, with N+1000 you bring in some UI experts to design a better interface that doesn't need to be trained in the first place.
There are two assumptions you can make based on this chart. The first is still more or less true today: as demand for a product goes up, supply increases, and price goes down. If the price gets too high, demand falls.
Er, no. I guess even smart people like Gates get simple economics wrong sometimes. It's understandable, though, I've always felt the P/Q axes should be switched, since people always speak about the price as the independent variable.
He's conflating the idea of pricing along the line with the demand line itself. Yes, at higher prices, less quantity is demanded and vice versa. Similarly, if you can sell the thing for a higher price, more people will be interested in and able to sell it, so the quantity available will increase, and vice versa.
However, "supply" and "demand" refer to the whole lines themselves, not points along the lines. Changing supply and demand means the lines shift. If "demand for a product goes up", the demand line shifts to the right, the equilibrium quantity increases, but so does the price, the opposite of what he's saying.
His misconception then leads to an even larger error in understanding with the next part:
The second assumption this chart makes is that the total cost of production increases as supply increases.
This is false. What it says is that with if the price you can get for a good is high there will be more suppliers. The canonical example is oil: when it sells for $20/barrel, only Saudi Arabia and places with it easily accessible will be able to profitably drill for it, and so the quantity supplied will be low. But at $120/barrel, you can now use fracking, deep water drilling, and other much more expensive processes, and so the quantity supplied will be higher.
Supply and demand is not about any individual producer producing more, and so it's not particularly useful in analyzing the behavior of a single firm. It's about the aggregate effect of multiple different parties, each with their own respective infrastructures and costs, profitably contributing to the supply (or not) at different price points.
1) Everything you say is correct, and I have no idea why you're being downvoted.
2) In my experience, the supply+demand curve is one of the most misunderstood concepts in economics, and it took me a long time in college until I found an economics professor who could explain it properly. A lot of really intelligent people I know don't understand it (even though they think they do), and in my experience most business people don't actually understand it (that's fine, they don't usually need to), but I am fairly shocked to see Gates misunderstand it... unless this article is written by an intern or something, which I find more likely.
Gates> The second assumption this chart makes is that the total cost of production increases as supply increases.
I think that's a poorly phrased segway.
What he actually seems to be talking about later is that marginal cost flattens after some number of units, but then continues at a fairly constant level (representative of material inputs).
Contrasting this with software and intangible production, where the marginal cost continues to decrease for much larger scales, before hitting a much lower floor (compute + disk + amortized development + maintenance).
And how the proportion of our economy governed by that different marginal cost vs volume curve has continued to grow. And that this has underappreciated implications on how we design economic law.
As he notes, Capitalism Without Capital is an exploration of this idea that he read and enjoyed.
Who exactly isn't paying attention? Gates never says. I don't know anyone who is reasonably educated in business/economics who doesn't understand the software/publishing business model, which is taught... literally everywhere. I seriously have no idea what he's arguing against.
Even with the supply curve, real-world commercial software virtually always has unit costs. It generally takes marketing and/or sales to acquire new customers, ongoing server and storage costs, customer support... and ongoing bugfixes and improvements merely to remain competitive and not fall behind in the market.
When Microsoft calculates the profitability of Word, their costs aren't "virtually free" or anywhere even close to that.
I was first annoyed at the amateur level of analysis in this piece. "Um, it's 2018, open source software runs production workflows in every single business; if you don't realize yet that every industry in the world is being reduced to becoming a commodity machine to run business rules & agreements as fluid software, then you're about 5 years behind the curve."
Then I realized it was written by Bill effing Gates. And that made me feel sad.
Firstly, because Bill is smarter than this. His lack of awareness of just how pressing this phenomenon is, indicates that he's either had his head in the sand, or the intellectual circles he swims in hasn't been ringing the alarm bells about this. (See my essay on the Changing Nature of Scarcity: https://medium.com/@pwang/the-changing-nature-of-scarcity-fc...)
Secondly, because of the irony. Bill Gates made his fortune having created a business around selling software. It would be good for him to review John Perry Barlow's "Selling Wine Without Bottles": https://www.eff.org/pages/selling-wine-without-bottles-econo...
As the world's first and most successful wine bottler, billg might find this perspective illuminating.
He seems to be arguing that policymakers aren't paying enough attention to this. He specifically calls out intellectual property laws, tax policy, and competition laws as three areas that have not yet adapted to this trend.
It sounds like your point is that the entire business world understands software in 2018. I don't think that's entirely true, but I'll accept that it's pretty close. However, I don't think Gates is talking about the business world.
Bill Gates is claiming that the old men who run the world have not yet internalized this. You are correct that this is taught in colleges now. We have to have a whole generation die off before that knowledge is widely applied.
92 comments
[ 3.4 ms ] story [ 103 ms ] thread"The portion of the world's economy that doesn't fit the old model just keeps getting larger." How can it get larger if no one is paying attention?
Or do you think the phrases mean the same thing?
(e.g. it doesn't matter if I have insufficient money or no money at all; life sucks in both of those alternate-universes)
They absolutely don't. I'd you have "not enough money" to meet your current monthly budget that means you can't afford to put anything aside for your vacation next year but can still get by.
If you have "no money", you can eat or pay your rent.
The two phrases have entirely different meanings, and usually with important differences in connotation.
He then says:
That has major implications for everything from tax law to economic policy to which cities thrive and which cities fall behind, but in general, the rules that govern the economy haven’t kept up. This is one of the biggest trends in the global economy that isn’t getting enough attention.
WTF, seriously?
https://www.imperial.ac.uk/business-school/knowledge/finance...
He’s just using the example from the book and doesn’t feel the need to restate it?
He probably just stuck with repeating what the book said or is just trying to keep it simple.
I suspect he gave this book review assignment to an intern that majored in something besides CS, business or Econ.
I think it is fine but confusing because it skips over some steps. A longer version might be:
... as demand for a product goes up, the demanded quantity goes up, which causes supply to increase (to meet the higher demanded quantity) which then pulls the price down (due to the increased size of the market and competition).
The net effect of this is decreased price overall due to economies of scale (since the total production volume of the product in the market is now higher).
They discuss it as if price is the independent variable and quantity (demanded or supplied) is the dependent, and yet any 9th grader would be marked down for putting their independent variable on the "y-axis".
It amazes me how something generally accepted as wrong stays that way.
How so? Econ major here, and I agree the axes should be swapped, but I don't follow your other point.
If I was a company director and the price of the commodity that I produced dropped by a lot, I would work harder to produce more of the commodity to offset the lower profit margins; that way I could earn the same bonus at the end of the year in order to make the payment for the mortgage on my yacht.
Also in the case of cryptocurrencies, it doesn't seem to work the way the graph suggests; when the price goes up, people buy more.
> Why is it that when the price drops, consumers would want to buy more and producers would necessarily want to produce less?
I would phrase it as when the price is lower, there are more consumers who would buy, and fewer producers to sell. To take the demand curve, for example, it doesn't explain one person's behavior at different prices. Rather, think of the curve as a thousand points, each representing a different person and their own maximum price at which they would buy the item. The supply curve is a thousand producers each with the minimum price they'd be willing to operate at.
> If I was a company director and the price of the commodity that I produced dropped by a lot, I would work harder to produce more of the commodity to offset the lower profit margin
Yes, this is already reflected in the supply curve. If your company is able and willing to produce that commodity at a lower price, you're already reflected on the supply curve as one of the points on the line below the current equilibrium price. You have what's called a "producer surplus" if you'd be willing to produce at a lower price than the equilibrium.
Take the supply of uber drivers for example. If the price of rides decreases, you may to some extent see drivers adjusting how much they drive, but the main effect is to remove drivers from the market who have better uses of their time, less efficient cars, etc. Similarly, if the price increases you'll see new uber drivers coming online driving their clunky SUVs or whatever.
Or, in the case of cryptocurrency, the supply follows this as well: at a higher price, less efficient miners or people in locations with higher utility prices will be able to profitably mine, and so the number of people supplying computation will increase.
> Also in the case of cryptocurrencies, it doesn't seem to work the way the graph suggests; when the price goes up, people buy more.
This, too, works with the supply and demand curves, when you understand them properly. The concept at work, here, is that changing supply or demand (i.e. think of the thousands of people that make up each curve shifting their preferences in the aggregate) represents shifts of the curve, not movements along it. Increased demand is the curve shifting up and/or to the right (it's the same thing).
So, when the price goes up, people see it in the news and have FOMO and so "demand increases", meaning the curve shifts to the right. If you look at the graph again, you'll see that means the equilibrium price increases, which is what's happening. That, in turn, drives more breathless speculation and aggregate demand increases again, pushing the curve further to the right, driving up the price some more.
Concept check: the supply curve is not shifting, here. Per my earlier paragraph, the increasing price may mean more miners come online, but that's already what the supply curve means. A shift in the supply curve would be something like a new breakthrough energy technology making everyone's utilities cheaper. That means if everyone's minimum price before was $x, they're now willing to supply computation at $x + 5, shifting the supply curve to the right and (if you check the diagram) driving down the equilibrium price (since the # of bitcoins in circulation will increase faster and the demanded fee by the miner's will be less).
The cylinder-with-a-piston used to teach Boyle's law in secondary-school physics is a familiar example: in the short term pressure is a function of volume; in the longer term volume is a function of (external) pressure.
Given that any Y-X chart is an X-Y chart if rotated 90 degrees and then flipped horizontally, is there any real issue here, other than the axes being the reverse of the standard statistical style?
I was under the impression the Gates Notes were written by him, wrong?
Software is the extreme end of scalability. One good lesser example is manufacturing vs repair. As mass production took off in the time it took to repair one item a worker could make ten new. Wasting some material on defectives made more sense than repairing them usually unless the cost of failure is extra high, inspections more expensive or a more cautious approach yielded better viable output. It is only when the expense of new is high that repair remains intrinsically viable (if you enjoy repairing things economics don't matter as it is its own reward).
If fixed per unit costs are high extra supply does little good - especially if it is durable. You won't be able to buy $100 for $99.
That also means it’s cheaper to develop. Software is so popular since it’s only requires a developer and a PC. The return/investment is a lot higher and faster than traditional hardware development. Cheap and fast is what everyone wants.
Of course investors are now realizing, in the case of self driving cars (or cars themselves with Tesla), that hardware still plays a role, and it’s not cheap and fast with instant riches.
20 years ago, biotech was supposed to be the hot thing after the internet. Instead it has been software services, focusing on service labor. Biotech is not cheap and fast enough.
Also, his statement implies that digital assets have no real-world financial value. In practice, even if the investment fails, the codebase of the company, body of active users, and any user data gathered will have some value.
"It" refers to "intangible investment" here--
1. It’s a sunk cost. If your investment doesn’t pan out, you don’t have physical assets like machinery that you can sell off to recoup some of your money.
2. It tends to create spillovers that can be taken advantage of by rival companies. Uber’s biggest strength is its network of drivers, but it’s not uncommon to meet an Uber driver who also picks up rides for Lyft.
3. It’s more scalable than a physical asset. After the initial expense of the first unit, products can be replicated ad infinitum for next to nothing.
4. It’s more likely to have valuable synergies with other intangible assets. Haskel and Westlake use the iPod as an example: it combined Apple’s MP3 protocol, miniaturized hard disk design, design skills, and licensing agreements with record labels.
Dunno if they're fallacies so much as oversimplifications. But the oversimplifications aren't really the point; they're just meant to reference the concepts and give a hyper-quick overview if you've never heard of them. The rest of the review remains a valid point if you substitute more realistic economic concepts. This is my pre-emptive reply to the inevitable dozens of posts arguing about the oversimplifications. None of it matters to the point the author wanted to make.
Note I say "valid", not correct. Whether economists are undervaluing intangible assets is a rich and interesting question that I have only vague opinions about personally. I'm just pointing out that the argument itself does not depend on the oversimplified economic concepts used to introduce the point.
> The first is still more or less true today: as demand for a product goes up, supply increases, and price goes down.
Classic econ teaches as demand increases, prices go up. The exception is if the supply curve is flat (for instance, 0 marginal cost).
No, they're far enough off of correct that I wouldn't even call them oversimplifications. The statement of the problem is wrong:
The second assumption this chart makes is that the total cost of production increases as supply increases.
That's simply not what the curve says at all. The curve isn't about any one producer making more of an item (as the Ford example given), it's about how producers with different costs are able to profitably add to the supply or not, based on the price they can get for the product.
The review might be fine otherwise, but there's no reason to throw supply & demand under the bus to motivate it. In fact, if you frame it correctly (say, technical innovations have enabled more people to more easily write software, thereby shifting the supply curve to the right and lowering prices), you can still meaningfully analyze these situations with those curves.
Yes - over time supplies tend to change due to demand, of course, but that's 'long run' stuff ... not in that chart. The author is kind of misinterpreting the chart.
2) It's not 'sunk cost' that's the issue, this is about 'fixed' vs. 'unit' cost. Software is all 'fixed' (i.e. setting up the production line) whereas most material things are mostly about 'unit' costs (i.e. the cost to make an item)
3) Software sales are a tiny fraction of the economy because almost nobody is selling software. People are selling 'services', basically products that 'do stuff using software'.
The 'key ingredient' is usually some type of business or experiential knowledge, not some algorithm.
4) Software development doesn't finish at v1, it's ongoing.
You can sell a big industrial motor or fork lift that might be part of the fixed costs of running a small factory.
"as demand for a product goes up, supply increases, and price goes down"
No, when demand goes up, the price will go up as well unless supply increases by the same amount. The price will only go down when supply increases and demand doesn't keep up with it.
"the tools many countries use to measure intangible assets are behind the times, so they’re getting an incomplete picture of the economy. The U.S. didn’t include software in GDP calculations until 1999. Even today, GDP doesn’t count investment in things like market research, branding, and training—intangible assets that companies are spending huge amounts of money on."
First of all let's skip the part about what the US did until 1999 because we're living in 2018 at the moment. And then the fact that market research, branding etc. is not included in GDP calculation has little to do with the digital age. There has always been market research and branding, not just now. One might argue that it should be included but that is really a separate topic.
"It took time for the investment world to embrace companies built on intangible assets. In the early days of Microsoft, I felt like I was explaining something completely foreign to people. Our business plan involved a different way of looking at assets than investors were used to. They couldn’t imagine what returns we would generate over the long term."
No doubt things were like that in the 80's. But just take a look at the most valuable Dow Jones funds now and it's clear that intangible assets are valuated well nowadays.
It absolutely is an equilibrium in normal economics: https://en.wikipedia.org/wiki/History_of_microeconomics
"demand for a product goes up, supply increases, and price goes down"
Assumption 2 from article:
"The second assumption this chart makes is that the total cost of production increases as supply increases. Imagine Ford releasing a new model of car. The first car costs a bit more to create, because you have to spend money designing and testing it. But each vehicle after that requires a certain amount of materials and labor. The tenth car you build costs the same to make as the 1000th car"'
In assumption 1 - he's not describing how equilibrium works at all. Equilibrium is the price the market will clear at given a supply and demand curve. If 'demand goes up' - given a fixed supply demand curve ... 'price goes up'. So what the author must be referring to is some kind of market response, i.e. if demand is increasing for something, more suppliers will come along and build it, create more competition, and drive prices down. This has little to do with the Supply/Demand curve.
I don't even know what assumption 2 is about, I get what he is saying but it has nothing to do with supply and demand curves.
Anyhow - even if there is logic in his statements, neither assumption is drawn from the Supply and Demand curve he is showing.
With assumption 2, he's saying that in software (and many other businesses now), supply curves do not necessarily slope upwards. For software, producing 100,000 units takes exactly as much money as producing 1,000 units, and can be offered at the same cost.
The author (who is Bill Gates, BTW, and can be assumed to know a thing about microeconomics) only alludes to this and doesn't say it outright, but there are a number of knock-on consequences to flat supply curves that explain much of the counterintuitive properties of the tech industry but aren't well-explained by classical economics. For example, the "lottery economy" aspect (where there is one company that ends up dominating a market and the size of that company is dictated only by the size of the market) comes from zero marginal costs: once a software firm has built a solution that satisfies a market, there is nothing stopping it from expanding to serve the whole market far faster than any competitor can enter. The existence of the VC industry follows from this, because when success is binary and can't be interfered with once discovered, it makes sense to spread capital across a number of bets, hoping for a hit, rather than rationally analyze how much additional return invested capital will give in an existing business.
Google & Facebook's dominance also stems from this zero-cost-of-production property. When your cost of production to provide the good or service is zero, then the only thing blocking your continued expansion is finding & convincing potential customers. With CAC forming an increasing percentage of your total marginal costs, any services that lower it or make it scale better can realize huge profits.
I'm not sure what you mean, this is 100% covered in general economic theory with barriers to entry.
> Google & Facebook's dominance also stems from this zero-cost-of-production property. When your cost of production to provide the good or service is zero
But it's actually not 0 or even close to it. There are still economies of scale at play, if Facebook were to lose millions of users their costs would grow per user, which is what a marginal cost curve represents. Just because the curve is different doesn't mean it doesn't exist.
I'm pretty disappointed with Gate's post, because every academic economist I know is mocking it...since it's not well represented or thought out. It ignores so many principles and theories in economics for the sake of being "ground breaking." Even though most of the theories are hundreds of years old and still apply completely.
On the side note, crypto currency is the new new intengible assets that is hard to internalize. Ironically, Gates has wrote them off as not worth calling them assets because they aren’t store of a value.
There's a shrink-wrapped cardboard package that contains Windows 98. It's the size of a cereal box, and at least as colorful.
Today, I'd be intensely annoyed if I had to wait for and pay for a large box to ship - I'd much prefer an instant download, but the box was much more tangible than an email with a download link and license key.
Was this not the reason software was distributed in CDs? I can understand MSFT/Apple using that packaging for customer experience.
The problem was that you couldn't convince consumers to pay for software downloaded over a modem. (The infrastructure wasn't in place anyway - nobody was going to send their credit card number unencrypted to a random BBS sysop, and encryption was outlawed as a munition.) Many software firms in fact had huge problems with piracy over modem, and would build copy-protection into their software so you needed the original physical disk to run the software.
The idea that you could pay for a license key that unlocks software you download over the Internet only happened in the early 2000s, driven by combination of remaining restrictions on encryption being dropped, easy-to-integrate payment solutions for webpages coming to market, and widespread broadband adoption meaning that most consumers could easily download a big file.
You didn't have this for business software, but you did need something to keep the disks in. I think the largest such thing I saw was 100+ floppy disks; here's Raymond Chen on floppy installs of Windows: https://blogs.msdn.microsoft.com/oldnewthing/20050819-10/?p=...
The user manuals were absolutely necessary for complex software. Most users had only a single low resolution monitor back then so online documentation had poor UX.
As someone with software and hardware at Ford in Dearborn and at many of their suppliers, this is not true. Ford knows that the costs decrease over time, and demands part of this cut from their suppliers. The contract includes a decreasing price over time or over quantity, because ostensibly you're increasing efficiency and writing off initial engineering costs.
I imagine that Microsft does the same with their hardware division, and agree with the sibling comments that Gates probably did not write all of this.
What hopefully will happen over time is that these rent seeking entities are squeezed out of the market over time, otherwise we are entering an area of corporate feudalism. This can be prevented as long as the assets needed for production are easily and cheaply available, which is the case thanks to the Free Software Movement.
This is quite similar to "free" (as in beer, let us not go down that tedious path). Say it only provides thousands of dollars of value and it's still pretty similar.
I don't think any economic theory is that naive.
It's actually the cost of production divided by the number of sales, where the cost of production can be many millions of dollars worth of development and testing and the size of the addressable market can in some cases be very few.
JetBrains has the best of both worlds with a fallback license to let you keep using an old version, or a subscription to get continued updates.
Games are becoming increasingly long lived pieces of software as well. With bugfixing and updates continuing for years. The industry is struggling to find novel ways to make money in that environment, which is why you get things like loot crates. I would prefer an honest subscription over that.
What I see is a devaluing of software as a product. I should be able to spend a couple of hundred hours on a package for a framework and sell it, but there is always someone offering a free version and then using weird and sometimes shady ways to make money. I wish it were more common to just sell code for money, instead of people pretending they are altruistic and then selling your data, selling your eyeballs or locking you into their ecosystem. Subscriptions are at least an open and honest transaction.
1. While the cost to produce N+1 and N+1000 version of the software is same in terms of producing the code, if you have a piece of software that functions more like a specialized tool, training user N+1 and N+1000 can be radically different and therefore more time consuming and costly.
2. Sunk cost of development: IP has a lot of value. If you work in producing custom solutions a lot of these solutions have components (code) that work across customer needs. Development may fail on one project, but helps in another. Overtime you have the components to sell a COTS
3. Goods actually aren't intangible, value is. Electricity costs money and it takes the infrastructure to support a small city to run a large-scale server farm. What's missing in the economic model are the costs we don't always factor in or more importantly that are publicly subsidized. Look at the breaks companies get at MSoft, Google, etc on utilities and land.
https://smile.amazon.com/Capitalism-without-Capital-Intangib...
Er, no. I guess even smart people like Gates get simple economics wrong sometimes. It's understandable, though, I've always felt the P/Q axes should be switched, since people always speak about the price as the independent variable.
He's conflating the idea of pricing along the line with the demand line itself. Yes, at higher prices, less quantity is demanded and vice versa. Similarly, if you can sell the thing for a higher price, more people will be interested in and able to sell it, so the quantity available will increase, and vice versa.
However, "supply" and "demand" refer to the whole lines themselves, not points along the lines. Changing supply and demand means the lines shift. If "demand for a product goes up", the demand line shifts to the right, the equilibrium quantity increases, but so does the price, the opposite of what he's saying.
His misconception then leads to an even larger error in understanding with the next part:
The second assumption this chart makes is that the total cost of production increases as supply increases.
This is false. What it says is that with if the price you can get for a good is high there will be more suppliers. The canonical example is oil: when it sells for $20/barrel, only Saudi Arabia and places with it easily accessible will be able to profitably drill for it, and so the quantity supplied will be low. But at $120/barrel, you can now use fracking, deep water drilling, and other much more expensive processes, and so the quantity supplied will be higher.
Supply and demand is not about any individual producer producing more, and so it's not particularly useful in analyzing the behavior of a single firm. It's about the aggregate effect of multiple different parties, each with their own respective infrastructures and costs, profitably contributing to the supply (or not) at different price points.
2) In my experience, the supply+demand curve is one of the most misunderstood concepts in economics, and it took me a long time in college until I found an economics professor who could explain it properly. A lot of really intelligent people I know don't understand it (even though they think they do), and in my experience most business people don't actually understand it (that's fine, they don't usually need to), but I am fairly shocked to see Gates misunderstand it... unless this article is written by an intern or something, which I find more likely.
https://en.wikipedia.org/wiki/Supply_and_demand
It does a good job of getting into and linking to its nuances, although it's not exactly a tutorial or anything.
I think that's a poorly phrased segway.
What he actually seems to be talking about later is that marginal cost flattens after some number of units, but then continues at a fairly constant level (representative of material inputs).
Contrasting this with software and intangible production, where the marginal cost continues to decrease for much larger scales, before hitting a much lower floor (compute + disk + amortized development + maintenance).
And how the proportion of our economy governed by that different marginal cost vs volume curve has continued to grow. And that this has underappreciated implications on how we design economic law.
As he notes, Capitalism Without Capital is an exploration of this idea that he read and enjoyed.
Even with the supply curve, real-world commercial software virtually always has unit costs. It generally takes marketing and/or sales to acquire new customers, ongoing server and storage costs, customer support... and ongoing bugfixes and improvements merely to remain competitive and not fall behind in the market.
When Microsoft calculates the profitability of Word, their costs aren't "virtually free" or anywhere even close to that.
Then I realized it was written by Bill effing Gates. And that made me feel sad.
Firstly, because Bill is smarter than this. His lack of awareness of just how pressing this phenomenon is, indicates that he's either had his head in the sand, or the intellectual circles he swims in hasn't been ringing the alarm bells about this. (See my essay on the Changing Nature of Scarcity: https://medium.com/@pwang/the-changing-nature-of-scarcity-fc...)
Secondly, because of the irony. Bill Gates made his fortune having created a business around selling software. It would be good for him to review John Perry Barlow's "Selling Wine Without Bottles": https://www.eff.org/pages/selling-wine-without-bottles-econo...
As the world's first and most successful wine bottler, billg might find this perspective illuminating.
From personal experience, this covers: any kind of software-driven hardware, health care, tax, CAD, POS, finance, transportation, & retail.
They're changing. But I'd estimate most Top 500 companies are at 0-25% open source migration of core business processes.
It sounds like your point is that the entire business world understands software in 2018. I don't think that's entirely true, but I'll accept that it's pretty close. However, I don't think Gates is talking about the business world.