> If you don't look at the price, it's the most stable and predictable service and protocol in the existence of the internet

Well, when you put it like that you just convinced me to put my savings into SMTP. Obviously you've identified the key features of a store of value.

Well, I've had BTC transactions drop out of the mempool (so, they were not executed, but you don't know that immediately, you only know days later), and transactions that worked took between a few minutes and more than half a day. The fees ranged from virtually nothing to around 50 USD. I certainly would not call that predictable and stable. I mean, you don't even know when the next block will arrive.

I would be suprised if a valid normal transaction with 50 USD would drop out of a mempool. What was the transaction weight? If you send the transaction data, I would be interested to look at it.

I'm usually overpaying (20 satoshis/byte, usually under $1 for a $2000 transaction), and I always get into the first block, but I never pay $50 fee.

The problem is Bitcoin Core may have (will have) unknown vulnerabilities and those might not get patched correctly once discovered.

Yes, this is true, I'm mostly worried about the elliptic curve signature part, as everything else could be fixed with an emergency hard fork (except SHA256).

Sadly OP_CAT operation is disabled (or substring equality operation), which would make lamport signatures available again for high value transactions. I would love it if lamport signatures would be enabled again (it would be quite easy to do), but I'm afraid that there isn't enough concensus to do it at this point, because some people would think that it's wasteful, and also lamport signatures are dangerous, as they can be used only once.