“Safe” comes down to your risk tolerance. Do you trust having all your transactions be public to the world forever and that the US law will protect you when you make a transaction that is “politically unpopular” by those with power? Or do you trust that US constitutional rights will be strong enough to not send you to prison for concealing where all your UTXOs comes from?

There are no right answers, but if the war on drugs has taught us anything, being on the side that favors a hydra typically wins in the long run. ie. As more privacy tools pop up, with fewer centralized architecture pieces, the harder it is to police whatever the policy might be at the time. Freedom will ultimately win if we have enough people willing to stand up for it.

Ultimately lightning will probably become the glue between different ways of “sharing” UTXOs, if all these different options can speak lightning they become interoperable allowing for instance payments with minimal on-chain fees. Just like the base layer isn’t for coffee payments, eventually neither will lightning. The fees associated with opening and closing channels becomes economically when the amount of transaction across those channels goes up a ton. If these channels are connecting lots of “shared” UTXOs solutions, they make a lot of sense. Right now it’s hard to see this as we’re still in the process of building out and deciding the best way to do “shared” UTXOs, cheapest way is just have an sql db and be a custodian (but is of course not ideal for all the reasons). fedimint is a step in the right direction which doesn’t require a base layer chain and gives added privacy and reduces rug pull risk vs having a single custodian, but there are still issues with that solution. There are other options in the pipeline that might be possible on the base layer, however, any changes to the base layer take time (which is a good thing). At the end of the day, the question becomes “are the fees I’m paying for opening + closing a channel worth the economic benefits I get from having it” which is a pretty reasonable incentive. Early days, lightning is still a child, at ~5 years old, compared to many of the tools we use in tech today.

I believe you’re correct, but I’d have to double check

No, since these are not manual closes but ones that occur due to the nodes thinking the other is lying

My only disagreement comes from force closures due to issues in lightning implementations not agreeing on a “fair” fee rate. These are expensive bugs whose cost are suffered by the node runners. To near no fault of their own, a massive expense occurs only due to higher fees in the base layer and a difference in how that propagates to the lightning node compared to their peer. The fix, of course, is to improve the implementations and how they come to an agreed upon base chain fee (many smart people are working on this). This clouds whatever benefits having higher fees on the base layer may proved in giving a clearer market of opening and closing channels/node runners tuning their setup with settings.