The central argument of my recent post 'A Future Without Fatalism' might be clarified via possible worlds analysis. If it's true that I will φ, that's just a claim about how the actual world is, i.e. how things will - as a contingent matter of fact - turn out. It is not yet to introduce any substantive necessities whatsoever, for something is 'necessary' just in case it is true in all (relevant) possible worlds. For example, we may define some future truth as 'causally necessary' just in case it is true at all those possible worlds that share our history and causal laws. Note that if causal determinism is false, then there will be many causally possible futures. That is, there are many possible worlds with pasts and causal laws identical to ours, and yet different futures. So we can say that each of those worlds are (causally) 'accessible' from ours at this time.
We might define various other accessibility relations. Compatibilists will want something a bit weaker than causal accessibility: they may instead define a world as compatibilist-accessible to an agent if that's a way things would have turned out had the agent's desires been suitably different. (Or something along those lines.)
Note that an agent can (in sense r) bring it about that P iff there is a P-world that is r-accessible to the agent. The free will debate may thus be understood as concerning which accessibility relation 'r' is relevant to moral responsibility and the "ought implies can" principle. I won't get into that here: even the relatively strict requirement of "causal accessibility" will do for my purposes. I merely require some non-trivial accessibility relation (i.e. which doesn't rule out in advance the possibility that multiple possible worlds may be 'accessible' to us).
With the framework thus set up, it should now be obvious why future truth does not imply any kind of (non-trivial) necessity or fatalism. For the question of whether we 'can' do something is a question about what worlds are accessible to us. But the fact that some future-directed proposition P is actually true does not suffice to determine non-trivial accessibility relations. So the following are entirely compossible: (i) P is true in the actual future, and (ii) P is false in some non-actual future which is nonetheless accessible to me at this time. That conjunction is just another way of saying that I can (but won't) make it the case that not-P.
So, again: the argument from mere future truth to fatalism is mistaken for the fundamental reason that it confuses the modal implications of 'will' and 'must'. This is made especially clear through the above possible worlds analysis, whereby 'will' claims are analysed merely in terms of truth at the actual world, whereas 'must' claims also concern accessibility relations. You can't get from one to the other without adding further premises. So mere future truth is no threat to genuine contingency.