Tuesday, December 9, 2008

A quick argument against Open Future views

Open Future views hold that if p is a proposition making a contingent claim about the future, then p is not true (on some versions, p lacks truth value, and on other versions, p is false).

Now suppose you find out that at t0, on an ordinary autumn morning in New York City, Bill Gates tossed down a million twenty-dollar bills from an airplane, each with a sticker attached saying: "Please take this. It's a gift from Bill Gates." You are then in a position to know the following fact:

  1. At t0+48hrs, at least one of the twenties is not be where it fell.

But if Open Future views are true, then you don't know this, unless you know something more about t0. For if t0 is somewhere in the last 48hrs, then (1) is a future contingent. Being a future contingent, you cannot know it. For the only propositions that can be known are ones that are true. But since you do know (1), Open Future views are false.

To put it differently: Oddly, if Open Future views hold, then whether you know (1) depends on whether t0 was in the last 48hrs, or further back, regardless of further evidence. Thus, what inferences can be made from a fact depends on how far back that fact is. This is not very plausible.

8 comments:

Jonathan D. Jacobs said...

Suppose instead of a pure open future view, one held a modified open future view, whereby the only future contingents that are not true are those that lack a present truthmaker. To the extent that there are present truthmakers for (1), say truths about the dispositions of various persons in the vicinity, then (1) is in fact presently true, even if t0 is not at least 48 hours prior.

Alternatively, one can suppose that while I don't know (1), I do know that, were a person to happen across one of those $20 bills, she would pick it up. (This is not an ungrounded counterfactual of freedom, but rather a grounded fact about her dispositions.)

Alexander R Pruss said...

But it is causally possible that no one picks up the money. Thus, each person might freely make the decision: "I need money, but someone else may need it more, so I'll leave it for others." It is vastly improbable that everyone would do that (it is, in fact, improbable for even one individual, except an exceptional one), but it is possible. Assuming that a truthmaker for p is something whose existence necessitates p, there is no present truthmaker.

Jonathan D. Jacobs said...

In which case the truth is that, say, were Suzy to happen upon it, the probability that she would pick it up would be x.

Is there a problem with the truth of such a counterfactual similar to the one we discussed regarding propositions that have both a probability but no truth value?

Alexander R Pruss said...

The counterfactual probability claims can perhaps be understood as conditional probabilities: "If A, then probably B" can be understood as claiming that P(B|A) is high.

Jonathan D. Jacobs said...

Are you suggesting that counterfactual probabilities can be reduced to probabilities? That seems wrong precisely because the counterfactual is not equivalent to the conditional. Or am I misunderstanding you?

Alexander R Pruss said...

Conditional probabilities are not probabilities of a conditional, nor are they a conditional containing a probability in the consequent.

Justin Capes said...

Alex,

Mightn't the proponent of the open view simply bite the bullet and reply that what we in fact know--given the truth of openism--is not (1) but rather something along the lines of

(2) It is very very (add as many "verys" as you think are needed) unlikely that any of the bills is where it originally fell.

Although, I am not particularly sympathetic to the open view, this does not seem counter-intuitive. Indeed, openists might contend that ordinary language doesn't always capture our precise meaning, and this is just such a case. Thus, they might contend that in uttering (1) most people mean to utter something more along the lines of (2).

Alexander R Pruss said...

Would the proponent of the open view make the same move about past cases where the probabilities are the same?

If not, that seems objectionable. If yes, that seems objectionable (because, after all, most of our information about the past is probabilistic in nature).