Mining is always 100% efficient. If you don't understand that, then you don't understand how a blockchain works. The purpose of mining is specifically to waste something of value. Without destroying value, there is nothing to keep the network anchored to 'truth'.

How much energy is needed to be burned? Precisely as much as is needed to combat 'spam' (both in the form of speculation and block-space consumption).

The only 'alternative' that has been pitched is the ironically named Proof of Stake. PoS has many, many problems (Andrew Poelstra itemizes many of them) not least of which is that they will degenerate into Work-based attacks/mining-strategies.

Thus far, miners have been reluctant to switch to an alternate chain. It is probably a very tall claim that this will ever happen at this point. SMTP was/is a terrible protocol, but the advantages of the network effect sustained its success. This effect is particularly pronounced in Bitcoin, mostly due to the way that blockchains work. The longest chain will almost always secure the most 'faith'.

Bitcoin's entire raison d'etre is that its transaction clearing mechanism is decentralized among untrusted parties. The best known way to enforce honest participation in this mechanism is by requiring workers (miners) to prove that they've expended resources (incurred a cost) in order to to be eligible to publish the next block and claim the block reward + transactions fees.

When a miner Finds a valid hash for a block, they prove that they've consumed a certain amount of electricity in a way they everyone can easily verify. In order for someone to perform a 51% attack, they have to use more electricity than everyone else, which is probably more expensive than just participating honestly. The security of the network is proportional to the amount of money it costs to attack it.

Mining is therefore neither wasteful nor inefficient, if it is understood that the goal is to have a decentralized, censorship-free currency that disincentivizes attacks.

Confirmation times are a feature, not a bug. 10 minutes secures a lot of energy in block, and drastically reduces forking. If you believe it is a bug, there are many solutions on the horizon, most of which create a market for settlement latency. Lightning network is the most promising.

I found confirmation time to be very fast. For small transactions (less than $100), it'd make less sense to confirm. For bigger transactions, it'd make sense to wait for 30 or 50 minutes.