1) I'm assuming you're including the block reward as part of the transaction "cost". That's not a valid comparison, since that is money injected into the network. Even if it was, upon reduction to negligibility of the block reward, that number will drop dramatically. The free market will eventually settle on what makes a fair transaction cost is in absence of a block reward.
Yes, when block reward is gone, that number may drop dramatically, but what then? what about 51% attack? Right now a PoW 51% attack on Bitcoin would cost hundreds of million dollars too, if the block reward is gone, then transaction fees better make up for it, otherwise a 51% attack would become very cheap.
Now if the transaction fee does make up for the block reward, then we are still back to to the original problem, hundreds of millions of dollars bleeding from the community each year, and paid to hardware vendor/electric company.
You're too short-sighted. If Bitcoin is still around by the time the block reward is insignificant to miners, then it's going to be worth a lot more than it is now, and a lot more transactions are going to be pushed through it than are pushed now, upping the reward per block. Transaction fees aside, think about it - the price only has to DOUBLE in 4 years to maintain the same value of reward to miners. In the last 4 years, the price has gone from $0.0025/bitcoin to $500/bitcoin. That's 200,000 times the current value. Now certainly, the value of Bitcoin cannot continue rising indefinitely, but one thing is for certain: if Bitcoin is successful, then block rewards will continue sustaining mining for decades to come. If Bitcoin ISN'T worth a lot more by the time the block reward is insignificant, then the Bitcoin experiment effectively failed, because the current value isn't enough for a significant number of people to be using it for transacting and storing value.
And you've still failed to convince me how money is "bleeding" from the community. What does that even mean? What do you expect miners would be doing with their money if mining wasn't available?
Finally, what's the correct level of security that Bitcoin needs? A 51% attack would be detrimental, but it wouldn't mean the end of Bitcoin. There are ways to counter and mitigate it, and the worst that would happen is a suspension of transactions until the attack is resolved. So, what's to say that $10M of attack prevention isn't enough? Currently it would cost somewhere in the hundreds of millions, but is that really necessary? I'd be interested to see some analysis in that regard.