How is Filecoin supposed to work?

I published this on Observable a while ago but in the interest of keeping all my content here for the long term, I'm syndicating it. Filecoin has neither replaced S3 nor died off in the time since publishing, so I consider it still fresh.

How is Filecoin supposed to work? Really, how is it supposed to work?

So as far as I understand it, FileCoin is trying to supplant Amazon S3 for file storage. It is extremely hard to see how this would work.

How a company might work

If we assume that the basic principles of economics stay the same, the basic economics of selling storage consist of:

1. The cost of providing the service: hard disks, networking, paying people to keep the hard disks spinning, buying new hard disks when they break, paying for electricity and air conditioning.
2. The income of people buying the service.

The margin between the two is profit.

File sharing is a competitive business

Traditionally, we understand file storage as a competitive market. Some players, like Amazon, have higher margins than others, but at the end of the day a company can compare S3, B2, Wasabi, R2, and any number of similar services with prices, and if those services are a better deal, then they'll buy those instead.

How FileCoin is different

FileCoin intermediates customer and provider in a way that makes providers "more" interchangeable and homogenous. You put FileCoins in, you get storage out, and in the middle you might choose a provider. But you aren't signing a contract with a business, and probably not paying taxes on anything.

In theory, FileCoin would create an efficient market through competition. This probably wouldn't be true immediately, might not be true ever, but let's accept the long-run theory.

Thought experiment one:

Okay, so how does this work? Here's a thought experiment:

I, Tom, want to compete with Amazon S3. I set up an LLC or C Corp, buy some hard disks, put up a website, accept payments. Does this work? Is this a common business plan? Generally, no, right? I don't have any scale advantages. I don't have a contract with a fast internet provider for discounted connectivity. I don't have a relationship with multiple hard disk providers like Backblaze does, or network colocation like Cloudflare or Amazon. I'd lose, right?

So conversely, I, Tom, become a node on the FileCoin/IPFS network. Does this work, and why? IPFS, as a distributed system, as far as I can tell, is strictly less efficient than a "trusted, centralized" system.

How are the coins supposed to be useful?

FileCoin uses its own unit of currency, the FIL. How does this work? The price changes through supply and demand, as well as, like other cryptocurrencies, extreme volatility, manipulation, and so on. The SEC is not there to catch the bad traders.

So, let's say the price of a FIL changes from $1 to $100. If I paid 1 FIL ($1) for 1 gigabyte of storage yesterday, would I pay 1 FIL ($100) today, and why? If I'm a buyer, is this price volatility a benefit in any way?

It's extremely unlikely that my business is charging for its widgets in units of Filecoins, but my costs are priced in Filecoins. What's the value of having a different currency for different things, with its own volatility? Isn't this mostly downside? Companies that have to deal with multiple currencies tend to spend a lot of time and money hedging that risk away. The risk isn't seen as a good thing: if your public company starts making exchange-rate bets, that's generally a sign that you've strayed from the thing your company was supposed to be doing in the first place.

The math

Here's my mental math with FileCoin:

• File storage is a low-margin, not-very-fun business with structural bars to new entrants
• FileCoin makes file storage cool because it adds an element of speculation and technological pizazz, but it doesn't change any of economics or structural problems
• This doesn't seem like it'll work.

Non-answers

Some non-answers here are:

1. FileCoin is not the end goal, it's just a jumping-off point, and the real prize is services built on top of the Protocol Labs stack, like the virtual machine. Sure, fine, not really an answer. And if the economics at the base layer don't work, how are we supposed to build on top of them? The foundational layer of economies tends to be things that are stable.
2. The tokens go up. As I wrote before, the tokens go up is fun for people who have tokens, but as far as I can tell, price volatility is purely a negative for the central thesis of buying and selling storage. Would token appreciation cause people to buy more storage, or less, or do we probably convert FIL to USD and spend the equivalent of USD? Even in a supposed crypto future in which Ethereum and Bitcoin are reserve currencies, we'd still convert from FIL to that unit and vice-versa, right?

Non-statements

• IPFS is neat, in some ways! I think it's not very useful yet and it's pretty over-architected and overpromised, but it's cool and cool people have worked on it and want it to be good. FileCoin is different from IPFS.

The slightly-expanded argument

Take a utility project - like HiveMapper, FileCoin, or Helium. Think about what it is without the crypto element: just one person selling wifi to another, or streetview images, or file storage. Is that thing high-margin?

We have comparables for this. We know the margin of big, scaled storage providers. Smaller ones, without magical differentiation, will have lower margins because they will have structural disadvantages (e.g. they'll pay retail for bandwidth). We know how much people get paid to drive for streetview competitors: it's not much.

Add in the crypto coin element, and figure out why it's a goldrush opportunity or even just a sustainable business, and figure out why. What's the thing that turns these low-margin grinds into such great opportunities?