If downloading was free everyone would be doing it

South Schodack , NY
New York US

Part two in a series of two or more of topics relevant to the Network Neutrality debate

I’m shocked that I need to explain this, but prices are determined by supply, demand, and sellers’ decisions about the best ways to optimize supply and demand. Over time they are determined by nothing more than these.

Are there production costs? Of course. But these are also governed by the balance of supply, demand, and sellers’ decisions about the best ways to optimize supply and demand. Wages are determined by workers’ willingness to work for wages versus the other options they have, materials costs are determined by the willingness of others to buy.

Are there other factors? No, not really. There are facts that complicate these equations, and those are important facts. One is the «stickiness» of prices, where the market prices take time to adjust to changes in supply or demand. Another is lack of transparency, where parts of transactions are hidden from participants in the voluntary trade of goods and services. That lack of transparency might be fraud, but isn’t necessarily.1

Furthermore all trade takes place between humans, and humans are sometimes irrational.2 They make choices based on faulty assumptions, bad information, and often without regard to consequences of their choices.

Supply, demand, and prices are linked to one another. When demand increases and supply stays the same, prices will tend to rise. If prices do not rise, there are shortages. Shortages are bad because the people who get the goods are the people who arrive first, not the people who are most willing to pay most for them.

Consider the differences between taxis, public transportation (like trains and buses) and ride-sharing services like Lyft or Über. During times of higher demand (such as rush hour, or before or after an event like a concert or a baseball game) the number of taxicabs does not increase, and neither does the number of buses or trains.3

During these times, the buses and trains become crowded to the point where riding them becomes increasingly unpleasant. Demand is higher than during the rest of the day. While at two o’clock you might be able to find a seat on a bus, at five-thirty the bus or train will be standing-room only, and commuters may be packed so closely together that they can’t move from the spot. It may not even be possible to board the bus.

Taxicabs similarly lack the ability to respond to increased demand, but with immutable prices the shortages take another form: lengthy wait times. A taxicab across town costs the same no matter what time of day it is, but at peak usage times you may have to wait an hour or more before one shows up.

With ridesharing services where variable prices are dependent on demand, a ride across town might cost two or three times more during peak times than other times. The increase in price turns off many people. The convenience of the ride is not worth the price to them. To others, people for whom getting across town is more important, more urgent, or simply who have more money (making the relative value of money less) the tradeoff is worthwhile. There is still some demand, but less demand in terms of the number of customers willing to pay that price. Therefore those who are willing to pay the higher price don’t have to suffer crowding or wait times.

Decreasing prices has the same effect, but in the opposite direction. More people will be willing to pay one dollar for an ice cream cone than will be willing to pay five dollars, assuming the quality of the ice cream is the same. If the shop with the five-dollar ice cream cones finds itself selling very few ice cream cones, they might drop their prices and find that they have to resupply faster to keep up with all the extra customers. If they drop the price to a nickel, even people who hadn’t planned to get ice cream will stop in because the price is such a low barrier.

The problem with low prices is that all the people who walk past the shop might come in and buy ice cream, causing the shop to run out of ice cream. Then the person who really wants a cone comes by and is turned away. Prices should be set so that the people who have access to a finite resource are the ones who want it the most.

Telephone companies used to all price their service based on the amount of time that was used. The longer the distance the call was, the higher the rate. This reflected the cost of manufacturing the network infrastructure over long distances.

During the early rise of the Internet, Internet Service Providers used to supply their connections with a finite number of hours, reflecting the cost of the phone lines and the network access they needed to provide those services. If you wanted more hours of connectivity, you had to pay more. As a result, that connection time was treated as an expensive resource. People checked their email only once a day, or connected to the ’net to do some specific tasks such as research, then disconnected. If one wanted to download a very large file, one might decide to obtain it some other way rather than spend the necessary time online.

When broadband became common, connections started to be always on, but at roughly the same price. People could now download or upload as much as they liked as often as they liked. Service providers may have assumed that usage would increase some, but they didn’t count on the amount in increase. Many of the services which are common today simply didn’t exist then.

The problem with flat-rate «unlimited» Internet service is that when the price goes to zero, demand will go to infinity. There was still a monthly price, but once that monthly price was set it cost the same whether you looked at your email once a day or once an hour. It was the same price whether you chatted with people on IRC at a few dozen bits per word, or had a voice conversation at a few thousand bits per second. In fact, it was the same price even if you made a video call at a few million bits per second.

Instead of just downloading low-bandwidth text files (eg web pages) people more often started to download music. After they started downloading music, they started downloading videos. And then they started streaming all of the television and movies that they watched instead of using an over-the-air antenna or cable TV connection. Watching a video online went from seeing a choppy postage-stamp sized image on a screen to ultra-sharp 4k video, the equivalent of sixty eight megapixel images every second.

Once bandwidth (after the initial base cost) became free, people set up webcams, even pointing at nothing in particular, just because they could. They stopped getting as angry when they went to a website and saw a video advertisement start to play automatically. They started storing their music and videos «in the cloud» meaning they downloaded their favorite album again and again, every time they listened to it. A decade ago these things would have been seen as comically wasteful. Two decades ago they were simply not possible.

But as the old saying goes, there’s no such thing as a free gigabyte. Someone is paying for that data to from place to place. The cost per gigabyte has been going down steadily, but it still requires equipment, and people to keep that equipment running. Promising unlimited access means the demand for that access will continue rising at an outrageous and irresponsible pace. The people responsible for providing that bandwidth aren’t the same ones deciding to use it.

This is not to say that it’s bad or wrong that these services have become available. The progress made has been wonderful. Digital services are abundant. But there’s a difference between abundance and actual limitless supply. So long as service providers promise supply without limit, demand will continue increasing. People will find ways to use bandwidth in ways we might find as ridiculous today as people found the idea of streaming movies twenty years ago. Just as our software gets more complicated to use up the resources of ever-more-capable processors so too will we use up whatever bandwidth is available as it is created.


  1. One should not forget externalities, where benefits or costs are spread outside the confines of a transaction. It’s not what this post is about, but you really shouldn’t forget about them. And when not forgetting about them, also don’t forget that attempts to leverage or ameliorate externalities will also be subject to laws of supply and demand and may create their own externalities. OK, I’m not forgetting that and neither should you. Let’s continue on. ↩︎

  2. Please note that «irrational» here does not mean «emotional». If a trade provides pleasure, such as money exchanged for a movie ticket, the trade can still be fully rational even if the benefit isn’t. Different people have different things they value. That’s a feature, not a bug. ↩︎

  3. This is not always true in real life. Most municipal bus and train schedules have more frequent stops and additional commute-time-only routes. But the increase in supply is usually a small fraction of the increase in demand during these times and the described effects are common. ↩︎