The way I'd put it is 3) In its current state, you can get string theory to "predict" just about anything. The reason is the enormous amount of string theory

The way I'd put it is 3) In its current state, you can get string theory to "predict" just about anything. The reason is the enormous amount of string theory vacua, each giving rise to different "laws" of low energy physics. You've probably seen the old estimate that 10^500 such vacua exist. That turned out to be a slight underestimate: https://arxiv.org/abs/1511.03209 https://arxiv.org/abs/1511.03209 Even if you try to narrow down your search to vacua with the same rough features as the Standard Model (types of particles and interactions) you end up with numbers like 10^723: https://arxiv.org/abs/1810.00444 https://arxiv.org/abs/1810.00444 (Disclaimer: what follows is a caricature meant to illustrate the magnitude of the problem, not a rigorous argument.) So in principle you can imagine performing lots of precision experiments to measure a bunch of quantities which can be derived from string theory, and then run through all those 10^723 vacua to see which ones match your data. It will take a while. Can you derive precise parameters of low energy physics from string theory vacua at a rate of one per second? The number of seconds since the big bang is less than 10^18. You do the math. And even if each quantity you measure can discriminate between two halves of the remaining search space, narrowing those 10^723 alternatives down to 1 will require more than 2400 such mutually independent quantities to be measured (10^723/2^2401 ~ 1). Alas, once you fix the rough features of the Standard Model, it has only about 25 free parameters. So, even given all the experimental data you need and eons to run your search, you will not be able to nail down one string theory vacuum as the one we are in (assuming that string theory is correct), if all you know about experimentally is the Standard Model. That only reduces the uncertainty about which string theory vacuum we are in to 10^723/2^25 ~ 10^715 different alternatives. You need to discover new physics at higher energy, to bring in more free parameters to use in your search. Something like 2376 (!) new free parameters. Only then will you be in a position to finally say that future discoveries in particle physics will be able to decisively rule out string theory (or not). Until then, whatever comes out of an experiment will just tell you that some vacua are out, but you will have plenty left.