(1) That's not made apparent in the sentence (why leave it for a footnote?), and (2) you haven't explained why that number is actually meaningful.
Edit: I don't know much about this stuff so please correct me if I'm wrong but wouldn't the cost of the "externalized price per transaction" be (price_to_mine-block_reward)/number_of_txs. price_to_mine and block_reward would approach each other over time.
The block reward will continue to half in regular intervals during the next decades, while the price of a Bitcoin can't continue to double in regular intervals during the next decades (see: limits of exponential growth, etc.).
This leaves us with two options:
1) Instead of being financed by the block reward miners will require higher mining fees in order to include transactions. With a decreasing block reward those mining fees will eventually approach the actual cost of a transaction (currently: $50).
2) There will be less miners so mining a single block will be cheaper. This will lead to a decreased security of the network and make the network more vulnerable to attacks.
Why don't you factor in the possibility of more transactions per block? The idea is that as the Bitcoin transaction volume climbs, the need for a block reward decreases.
Secondly, you're assuming that the mining pool is currently constrained by what is profitable. From what I have read about large mining operations, they are significantly into the green on profit margin right now, so there seems to be room for a decrease in profitability without losing significant portions of the mining network. Every time the mining difficulty changes this is tested already so we know that there is at least some wiggle room in the mining profit margins.
You mean that the number of transactions grows faster than the mining capacity? I've covered a similar case in https://news.ycombinator.com/item?id=7846564 (it's #2). In general, if you want more transactions you also need a better security, which leads to increased mining costs.
That does not seem like a similar case to me. You specifically say "less miners" in #2. Plus what do you mean by mining capacity? If you mean actual GH/s then that's not a meaningful measurement because better hardware in the future will give higher mining efficiency at the same mining cost.
It just seems like you're oversimplifying a really complex equation with lost of potential outcomes. Perhaps this short comment style just isn't allowing you to explain everything. Have you written a blog or anything that details your thoughts on this?
Edit: I don't know much about this stuff so please correct me if I'm wrong but wouldn't the cost of the "externalized price per transaction" be (price_to_mine-block_reward)/number_of_txs. price_to_mine and block_reward would approach each other over time.