Yea — he’s kind of confusing, but he’s my favorite..
Every minimally acceptable moral theory must, in some way, force someone who wants to be morally good to take the concerns of others into account. This is because, as beings that are also animal in nature, we can be selfish. We are only able to see the world from our own limited perspective and if we declared “moral” whatever seemed good for us at the time, it would be a disaster — especially if others followed the same rules as we propose to follow.
With the universal formulation, Kant is primarily making an argument using logic that forces us to consider the thoughts and thus the autonomy of other people. The big idea is that if you WANT to do something, you must not be doing a thing that involves the deception (thus the use) of others. It’s a call for imagined transparency. Imagine that whomever you are dealing with could read your true motives and plans — would they cooperate in your action? If so, your action passes the logical test of the universal formulation.
The desirability test is the second part — would you WANT to live in a world where others are acting the way you propose to act? It’s much simpler and not, in my opinion, the primary test of the universal formulation..
Notice something important about the formulation, which cleaned up says — act only on those maxims you would — as you are acting — will to become universal law. This theory only eliminates possible actions, it does not suggest the absolute “right” thing to do. In essence, the categorical imperative is akin to a veto — not a means for suggesting a right action or for choosing among right actions.
The “means ends” formulation is specific to interpersonal relationships, but it’s coherent with the universal formulation in that it is a specific instance of it — you may not “use” others or yourself as a means only, you must consider their ends. In other words, don’t violate the autonomy of others by treating them as a means or method to get what you want. Instead, respect their autonomy by letting them make their own decisions about their actions — based on full information. What it’s doing is requiring you to inform the other person in the interaction of your actual motives — so, instead of just imagining whether or not they’d assist you if they knew what you know (like the universal formulation), you must actually find out — by being truthful with them and then letting them make an informed choice.
Something that has always puzzled me about the means/ends formulation is the clause “yourself or others” — the implication being that you can use yourself as a means only. Isn’t it always the case that you are considering your own ends? I’m not sure how you can avoid it — if you want to be “used” (see the weird craig’s list ads, for example), then isn’t that your choice? If you want to choose to have limited information from which to make a decision, aren’t you also choosing that?