Deploying a whole ass web application on small, immutable smart contracts seems inane to me. Reminds me of something programmers do not because its powerful, but because its hard and it makes them feel cool.
If a Pi solves your use case, for God's sake don't use the EVM.
If a Pi doesn't solve your use case, because {particular reason}, then consider the EVM as an option.
Cryptography is the basis of Web 3.0, so in a formal sense there is none.
But I think that you rather want an example that is a little farther away from finance markets. Consider, for example, the (in my observation mostly dead) Namecoin:
> https://en.wikipedia.org/wiki/Namecoin
which attempts/attempted to implement domain names via a blockchain and is sometimes considered a precursor to NFTs.
It's a potemkin village.
No, not necessarily. For one let's qualify that statement. Smart contracts on Ethereum are executed 100,000+ times by every node on the network, which is providing security and uptime for the network.
Some software transactions have external costs associated with them, such as financial transactions. Doing an international bank transfer may simply boil down to a credit in one SQL database and a debit in another SQL database but it still can cost dozens to thousands of dollars depending on the amount, and the situation. Even a VISA credit card transaction costs 1.5-3%. You as a consumer don't see that cost, the retailer is eating it though, and marking up their products and services to accommodate that fee.
But ETH transaction fees do not scale with the value of the transaction, it simply scales with the demand for the limited block space.
Ethereum community has been actively developing solutions for this problem, by developing optimistic rollups and zero knowledge rollups, which are effectively able to compress transactions and scale the network throughout by 100-1000x. Innovations like this will get Ethereum to Visa scale plus.
This technology is still early, but rapidly improving. Currently deployed rollups are in the 10-100x scale improvement range.
Some rollups currently in production:
https://arbitrum.io/ https://www.optimism.io/ https://starkware.co/
On Ethereum, you can execute a swap from one currency to another (0.3% fee) and have the transfer execute to anyone in the world in 15 seconds. Even in a highly congested time on the network, that will cost at most a few dozen dollars. Doesn't matter if you're sending $1K or $1B.
Crypto isn't ready for retail use cases yet, but it's within the next few years (less than 3) that it gets there. Retail transactions will occur on rollups. The retailers will pay transaction fees of a few pennies to a few dozen cents. It will not be percentage based on the transaction value as VISA, AMEX, etc currently get away with.
For example: CryptoKitties. The only things in the on-chain state of CryptoKitties' contracts, are the kitties' ownerships, and their "genes" (an entropy blob, a bit like Pokemon https://bulbapedia.bulbagarden.net/wiki/Individual_values.) Because these are the two things that, if they were entirely in the hands of CryptoKitties themselves, you'd be suspicious that they might "cheat" on, by e.g. making their friends special kitties with genes nobody else can have; or deciding one day that kitties that aren't "cared for" should be forcibly transferred back to an "adoption agency." If either of these things were found to be possible, the value of kitties would quickly drop to zero.
Meanwhile, what exactly a kitty's genes translate to in terms of actual kitty features, is dictated by some arbitrary centralized Javascript on the CryptoKitties website. Nobody distrusts CryptoKitties on this aspect, because it doesn't really impact the value of their investment relative to other people's. If the same gene makes kitties render with wings one day and vampire fangs the next, that might upset people, but it won't make them do a mass sell-off of their kitties. After all, CryptoKitties could always change it back, so why would you sell "early" rather than wait to see if they do?
Or, in short: you put data in a smart contract if a game-theoretic prerequisite of your system working out is for certain data to be only updated in an openly-audited, multilaterally-agreed-upon way. (Good non-crypto example: the dealer's shuffle between hands at a card game at a casino. Done right in front of all the players, because otherwise they wouldn't trust it.)