This means the protocol can accommodate an increase of an extra 53-58% in txs (and that includes dust and spam txs like the ones we have now).
This is faulty thinking. We will start to experience serious issues well before 100% and user experience issues even before that. Already several times we have had two full blocks in a row and possibly other times more that I don't recall.
It is not enough that transactions occur eventually, they should also happen in a timely manner.
My postal service has at least 3 different speeds. A priority, B priority and express. And I can also use private courier for even faster. So 4 types of fees for 4 different delivery speeds.
If I don't care about the speed, I'll go second priority. If I do, I'll pay the premium. It's that simple.
There's no entitlement that I can produce all kind of spam and they have to be included in one or a few blocks (for peanuts).
You pay the fee, you are in. In a timely manner.
If you're buying a coffee or are a retailer with a POS terminal, you need the transaction to go through fast. If you're buying a tanker full of North Atlantic crude, you can wait a few hours. It's the small transactions that need to go through fast, and the 1MB cap is making that impossible. BUT You will never get the crude transaction if you don't get enough of the smaller transactions for Bitcoin to build up it's market cap and liquidity. There will be a need for LN type sidechains in the future ONLY if we don't get too greedy by building a xaction fee market before the BTC market is ready.
If you estimate how many MW/sec the mining network consumes and compare it to how many xactions/sec it processes, you can tell mathematically how much a BTC xaction really costs at a given power cost. Currently, this is between $2-5 PER transaction (according to Toomim, an electrical engineer) and is not significantly cheaper than legacy payment systems. This can be fixed by increasing the block size.