Is Filecoin a $257 million Ponzi scheme? [pdf]

Files and directories are just one particular data structure you can store on IPFS. Think of it as a global unified content-addressed graph database really, and Filecoin can store any arbitrary (sub)graphs within that.

The analysis rests on this ratio calculated for Airbnb: "Airbnb is valued today at approximately $31 billion while holding around 3 million listings in total. The average apartment in the United States was 934 square feet in 2016. In a hypothetical scenario, Airbnb is therefore valued at $11.06 per square feet. If one compares this number to the median price per square foot in the United States, which is $123, the Airbnb ecosystem diminishes the median price by a factor of 11.12."

I get this is back of the envelope so let's ignore the international dimension. The critical error in these calculations is the apples-to-apples comparison of the square feet of an Airbnb listing to a purchased square foot.

Basically - when someone pays $123 per square foot (median price in the U.S.) they are paying for the use of that space for an entire year. Said another way, they are paying $0.34 per day per square foot. Airbnb listings are not in general year-long, which means the same square foot of space is rented out over and over again throughout the year, adding revenues along the way. The 11.12 scaling factor is therefore off and too high. This leads the authors to apply an erroneous scaling factor to the Dropbox sizing to come up with an invalid valuation assessment for Filecoin.

It looks like a good average stay for an Airbnb property is about 5 days (https://blog.atairbnb.com/economic-impact-airbnb/). Let's also assume 50% occupancy rate (https://www.mashvisor.com/blog/what-airbnb-occupancy-rate-ca...).

That means an Airbnb listing is rented out about 37 times a year (365 / 5 days * 50%). The $11.06 per square foot times 37 revenue-generating occasions is $409.22 per square foot ($1.12/day/sq ft v. $0.34/day/sq ft for a purchase). So the correct apples-to-apples comparison here is to say that Airbnb increases the median price of the asset by a factor of 3.32. Aka Filecoin would need to deliver 2,580 petabytes - not 95,185. That's only 5x Dropbox, which is a minnow compared to S3 etc.

Why would that be? Most directly, because Airbnb allows the owners of the asset (sq ft in this case) to ask for a higher per day charge since the length of stay is shorter and thus the owner of the asset has more uncertainty over utilization - it's part of why 30 nights in a hotel costs much more than a month of rent. This is a well-understood economic dynamic and one that would apply to Filecoin storage miners given how miners using Filecoin could theoretically 'rent out' the same gigabyte over and over again to different buyers throughout the year - but also cover the risk of less than 100% annual utilization rates of their fixed-cost asset.

Honestly, reading this article made me more confident in the future of Filecoin. The authors highlight multiple times the better technical features than competition, proven dev team, and unwittingly call out why economically speaking owners of gigabytes under Filecoin could expect a higher return than if you were going to buy the gigabytes yourself for your own use.

Thoughts?

A crappy computer, no; a crappy server, yes. The available network is part of a server. A URL that points to a machine that can't be accessed whenever the owner goes into a tunnel, is a URL nobody will link to.

(Another part of the evaluation of "a server" is the available concurrent I/O throughput. There is a reason the term "slashdotted" exists, and thinking an iPhone—or even a home PC—makes a good server for serving anything with any exponential-interest-curve potential is it. This is also the reason why nobody has built the sequel to Iridium using cube-sats.)

> Unless this comes as completely free AND it comes built into the device

Yes, that's exactly the future that cloudlet advocates like IBM are talking about:

1. the ISPs run compute clusters at the edge, including both setting up traditional co-los, and making their in-home gateway modems into compute hosts as well;

2. the device/"cloud" providers (Apple, Google, etc.) make agreements with said ISPs to pay for time on those edge-clouds;

3. the device/"cloud" providers resell (possibly at a loss, to encourage lock-in) the use of the aggregate edge-cloud to the app developers on their platforms; or, they let developers use the edge-cloud in their apps for free, and pass the costs onto only heavy users, the same way they do with document storage today;

4. the users of these devices suddenly have cloudlet-enabled apps, and most don't have to pay a cent more.

Amusingly, you can already see that the device/"cloud" providers are trying to short-circuit this whole process, skipping the ISP step, by simply selling people consumer-electronics devices that could sit in their homes, connected to ZeroTier-like overlay networks, hosting mobile cloudlets. You know, like the Google Home and Amazon Echo.