Regardless of blocksize the block times are still 10 minutes which means you need one hour for the transaction to settle, per the 6 block standard. You can't compare that with LN.
It lets one party in a channel send arbitrarily large numbers of payments ("arbitrary" subject to the total payment value being capped to the channel capacity, as well as to the Bitcoin divisibility) to another without increasing the UTXO set as you would on-chain.
No that is not true. Each Lightning Channel needs UTXOs to be created before it can be used at all.
Its goal is rather to do trustless multi-hop off-chain payments on pre-opened channels.
Notice that Satoshi himself implemented simple payment channels in Bitcoin. (Without the multi-hop part)
It is all but clear that the much more complex LN contraption will help in the real world.
Furthermore, a case can be made that this going closer to decoupling Bitcoin from the economy on top. The "gold standard" argument hinted at above might be a too close historical parallel, really.
Because, eventually, the gold standard got abolished.
It's not a "terrible idea" it is what should have existed in the first place in the oficial blockchain
They're just adding a 2nd layer on top, calling it official and then saying "bitcoin is working"
Bitcoin is extremely early stage technology. It'll take years for a bunch of scaling developments and blocksize is just a temporary measure.
no need to be genius to just increase blocksize 1000x and "assume" gigabit connection which no one has.
Lightning network is a great idea, in theory, in practice however it will just not work for the problem you mentioned - no one is interested in providing this kind of liquidity.
As of why need blockchain at all - is because 1st layer is the court and guarantees you to own money from the channel. You just don't have to ask court for every coffee payment (i.e. broadcast to every single laptop in the world).
If the Lightning network is so great, why do we need a blockchain at all? If not, why not just increase the blocksize? It achieves nothing that couldn't be done on-chain. There have been successful tests of blocksizes of up to 1 GB or 1000x the current size. Even getting it up to a small multiple of the current size would dramatically cut transaction costs and waiting times.