How secure does the Bitcoin network need to be? What mechanism exists to determine that?

Rhetorical question, BTW. There is no market mechanism that determines what is sufficient security. This should become clear if one looks at what the network is choosing to pay the miners who secure transactions.

The block reward is now 12.6% of what it originally was, and tx fees have certainly not gone up correspondingly. So is the network now undersecured? Or was it previously overpaying for security? Again, rhetorical questions - the users of the network aren't paying for security, they're paying to have their transactions processed and have no choice but to subsidize the miners with the block reward as well. As far as how much security is needed goes, the Bitcoin answer is simply "more is more."

But in the future, as the block reward dwindles, won't market forces start to price in security? I'm sceptical, personally, because the only gauge of whether or not the network is paying its miners enough is whether or not the network works. So if, in the process of finding the proper price of security, the market goes too low, attacks will be possible and the price of BTC will likely suffer as a consequence. This would make it less attractive to mine BTC and less attractive to use BTC. You can see how this could lead to a downward spiral.

This is a very good point. At some point in the future the mining power will need to be paid majorly by the transaction fees. This depends a lot on how high the price can rise, but lets assume gold market cap. Then it would be maybe 3-4 halvings = 12-15 years until the share of block rewards in miners compensation will decline and the transaction fees have to compensate for it. In my view this is the biggest long term security and usability issue because transactions will probably cost hundreds to thousands of current day dollars assuming steady or slightly rising power usage = tech-innovation adjusted hash-rate.

Hundreds of dollars are too much for the average joe to regularly open and close lighting channels, so we might see some kind of centralisation.

The big question is if bitcoin can provide enough value as a international settlement backbone to fund the biggest pool of sha-mining hardware in the world. If an alternstive usecase for sha-miners comes up (be it blockchain related or not) that would also be a bigg security blow (but I don‘t see this on the horizon for now)

Answers to those questions are not yet known. The price competition between different cryptocurrencies should be the necessary market mechanism. Someone has to pay the price of mining, and if mining is not necessary, then it's not going to be paid. If it can be proven that the same security can be achieved with less mining, then either the Bitcoin protocol is updated, or a Bitcoin fork will become more valuable (and that fork will be named Bitcoin).

There are lots of cryptocurrencies which have infinite supply, which means that the mining rewards will keep the network running. For example Ethereum, Monero, Dogecoin and Grin. Bitcoin will rely on transaction costs, but it doesn't provide fully deterministic security as far as I know.

I don't quite see how price competition between cryptos is a mechanism that optimizes security spending, unless you mean that cryptos that fail security will crash and burn.

As for mining, it is necessary for reasons that have nothing to do with security. If you want to send a tx, you will pay a fee or perhaps hope for a kind miner to include your spend in a block. In no way are you signaling how much you think the Bitcoin network should spend on security.

Very interesting food for thought. Thanks.

This is a completely bogus analogy. Even if I accepted that bitcoin was analogous to gold (hint: it's not), the government secures Ft. Knox to whatever level of security it feels appropriate, and then it's done/maintained at that level.

It's not like Russia adds one more guard to their gold vault, the US has to keep pace or else if Russia has 51% more guards outside their vault, then by magic they can suddenly steal/double-spend gold.

Do bitcoin folks really not understand this, or are they all just pretending to be blind because they hope it will make them rich?

A country cannot hold a significant amount of cash/gold/value without spending the appropriate amount on military defence to deter other countries from seizing the value. The armed guards at Ft. Knox is a small amount of the total expense, the aircraft carriers also participate at least in part.

The army isn't just protecting the gold. It's protecting the country and everything that's in it, including the gold of course.