The study specifically selected people with no substance abuse problem.. if anything the author of the study were wary of what you say is a reactionary perspective.

True, and funnily enough french people don't use yearly gross. Most of the time they use monthly net, and, in the context of salary negotiations, will specify over how many months. E.g. "2000 net sur 13 mois"

Imho, the java aspect of it matters much less than the backend concepts. Are you already familiar with those ?

There's quite a lot of stuff to learn on the backend and it really depends on which layer you want to focus on. If you're interested in developing business services then i would recommend writing a restful API (for example for a to-do app) with spring boot and your preferred flavor of SQL database. That covers a lot already. From there you can look at how to scale performance up (caching, queuing, asynchronous).

Is it really that surprising that planes are cheaper than trains ? Rails require massive investments and upkeep cost. It's also much more difficult to run different operators on the infrastructure so there is very little competition. Not sure a tax is going to change any of that.

Good discussion topic! I recently went through a similar exercise and my conclusion is that the HSA scenario financially wins in all cases as long as my OOP cost remain below 90% of the max on average. In my case both plans had low/no premiums.

There is one major caveat though: The advantage the HSA plan has comes from tax savings on the pre-tax contributions and from the tax free growth of the HSA. This only works if you do not spend any of your HSA funds and instead invest them, paying for healthcare with post tax money. Meaning you need to have 1-2x your max OOP set aside at all time to be safe.

The HSA is a very advantageous account for tax saving but I think it's fundamentally flawed for actual healthcare costs because no one can predict what their costs will be next year.

I think a better approach is to assume your costs can be anywhere from 0 to the max OOP and see how the numbers work out for different scenarios. Then you can make a decision based on financial (not health) risk tolerance.