plus/minus epsilon

I used to work on the Crypto Team at my company, and while I was there I got introduced to a lot of different crypto projects and asked to evaluate if they were something it made sense for us to support. The answer was usually no.

The problem is that a lot of them fell into the pattern of "X, but (decentralized|private)" where X is something that already exists and is well-established. DNS, but decentralized. Social networking, but private. Storage, but decentralized. A CDN, but decentralized and private.

That doesn't work, because the only thing your project is ever going to provide better than X is decentralization (with a blockchain) or privacy (with encryption), and people en masse genuinely don't care about either of those. If they did, data aggregators like Google and Facebook wouldn't continue to grow. If they did, Internet infrastructure wouldn't be continuously more centralized.

Consider the recent Verkada breach where a surveillance camera company was hacked, giving attackers access to the live video streams of all of their customers. People's immediate reaction is "What do you mean, these big companies can't run their own surveillance cameras? How could they be okay with sending all their video through some random company?" The answer is that they can run their own cameras. They chose not to because it's hard and they care more about reliability and ease-of-use than privacy.

After all, Verkada is just the business version of popular Nest cameras, which are sold to consumers and work the exact same way. At least Verkada cameras are only in offices. Nest cameras are in people's houses streaming to the cloud all day, awaiting the same fate as Verkada, and the people that buy them don't care.

Focusing a project on decentralization and privacy is containing it to a niche, and it makes crypto a handicap instead of an asset. The business opportunity for crypto is not rebuilding things that already exist, but with a handicap. The opportunity is in using this technology to build products that are genuinely better, by the conventional metrics that consumers care about, than anybody else could build.

So to continue with the camera example: what are the conventional metrics that people care about when they buy surveillance cameras? Probably something along the lines of:

1. Reliability.
2. Ease-of-use.
3. Price.

And there are roughly two different classes of cameras you can buy: 1.) more traditional self-hosted IP cameras, or 2.) cloud-hosted cameras like Verkada and Nest. Cloud-hosted cameras win far and away on reliability and ease-of-use -- you just plug them in and connect them to the Internet! But do they also win on price?

• Verkada's cheapest camera costs $599 upfront plus $199/yr.
• Nest's cheapest camera costs $130 upfront plus $72/yr.
• UniFi's cheapest camera is $179 and requires a $299 appliance.

Assuming a 10 year lifecycle, Nest and Verkada are 2 and 5 times more expensive than UniFi, respectively -- but that just goes to show how valuable it is to their customers that they don't have to manage anything.

So how would crypto make this better? We probably can't make reliability or ease-of-use much better than cloud cameras, but it would be nice to move back to the cheaper fixed pricing model of UniFi. To do that, we'd need to eliminate all of our own monthly cloud costs, the largest of which is storage.

Nest stores short "clips of interest" for their customers, for the past 30 days. If customers have 2 hours of clips per day, and video is stored in 1080p, that means each device has about 18GB of video that we need to store. For about $20 we can add 64GB of SSD storage to the camera. But rather than have the cameras store their video themselves, we have them participate in a hybrid FileCoin-type network where each camera stores the video of others in exchange for others to store their video. The problems this would solve are:

1. Having the cameras redistribute video among themselves prevents data loss when one camera is stolen or damaged.
2. It allows video to still be viewed when the camera that recorded it is offline.
3. The service provider's operational costs are fixed, instead of scaling with the number of customers, which is what makes upfront pricing possible.

Assuming Nest currently sells their cameras at-cost, or at a slight loss, we should be able to profitably sell cameras with all the same functionality for $200 and no monthly subscription fee.

Note that while this product relies on a decentralized cryptographic network, and the design decisions that will be made while building this network will emphasize security and privacy, cryptography isn't the hero. The product is worth buying because of its better price, and cryptography is just the tool that would enable that price. The fact that the product is more secure, private, decentralized are a strategy credit: nice auxiliary benefits of achieving the goal this way, rather than through other financial or technical means.