Saturday, 16 May 2015

Unknown Unknowns and Omniscience

So a few years ago I made a video cheekily suggesting that Donald Rumsfeld's much derided speech on Iraq and weapons of mass destruction actually provide grounds for logically defeating omniscience (cheeky given that, as a religious man, I doubt he would have wanted his words used in this way).

If you are interested, here is the video:
The Impossibility of Omniscience: How Donald Rumsfeld Disproves God!

So yesterday I received a criticism of the argument made in the video via 'PasteBin' and, not knowing how the hell you respond to something on PasteBin - ok, being honest, not knowing ANYTHING about PasteBin - I thought I would address it here.

First off thanks to the writer for the message (which I will keep anonymous in case they wish to not be associated with it (unlikely) or me (more understandable!) )

Here it is:
  1. So Noel, I recall an older video of yours where you argued that Rumsfeld's "unknown unknowns" concept (not technically his concept, but that's a tangent) means omniscience is impossible. Essentially, I think your case could be summarized as follows (correct me if I'm wrong):
  2. -Omniscience is defined as possessing all knowledge
  3. -Unknown unknowns are propositions that an agent does not even know that they don't know
  4. -Known unknowns are propositions that an agent knows that they don't know
  5. -So, you say that omniscience is an incoherent property because it would require an agent know the status of the set of that agent's unknown unknowns.
  6. -And since the status of an agent's unknown unknowns is unknowable, this prevents that agent from knowing if there are other things that agent doesn't know.
  7. Assuming I've roughly characterized your argument against omniscience correctly, here's why I think it fails:
  8. As you'll note if you look into it a bit, omniscience isn't always defined as knowing absolutely everything. A common definition used, which still seems reasonable to call omniscience, is "to know everything that can be known" or else "knowing all possible things". This definition of omniscience seems to completely evade your criticism, as it is not predicated on a being knowing everything, and hence unknown unknowns are irrelevant, as to meet the definition of omniscience, the being doesn't need to know the status of its unknown unknowns, the set of their unknown unknowns merely has to be empty, save for the status of that set.
  9. And it seems that all the apologist has to do at that point is to point this out, and then rightly say that as an omniscience being is metaphysically possible, they can make use of the concept of omniscience.
  10. Thoughts?

I am sure you will agree a reasonable point and a reasonable summary of my argument.

So one thing I got wrong on that video and, to my detriment, never got round to refining was the part:
-So, you say that omniscience is an incoherent property because it would require an agent know the status of the set of that agent's unknown unknowns

So what was wrong with that line?
It seems to me that what I wanted to refer to was the contents of the set 'unknown unknowns' rather than the status of the set. In fact the status of the set IS known which is that it always contains at least one item, which is itself (a bit of a Russell's paradox, in fact, if you phrase it the right way).
So I prefer to think of the set as an intrinsically unavoidable known unknown of unknown unknowns.

So the status of the set may be said to be known in that we can be sure it is not empty (even if it only happened to contain the trivial self-referential item #the number of unknown unknowns) but we can't say very much else about it. As a result I'd rather think of omniscience as falsified on the grounds of both this singular known unknowns' existence and also by way of considering the nature of the potential things that could constitute unknwon unknowns.

I think it is the second of those grounds that is of interest here.

First let me say, I absolutely agree with the questioner that it is important to consider how we define omniscience and that we ought not get too hung up on extreme definitions. In fact, one of the most fascinating exchanges I have had on YouTube, long before the A+ drama and the end of the happy times, was with a guy called Michael who had the channel "RationalRoundTable". Michael was a very open-minded Christian and we had a good exchange discussing the kinds of things that omniscience coud minimally entail. I gave a Bostrum-like simulation scenario and it was interesting to discuss, with an Abrahamic monotheist, the possibility that even something such as the ability to potentially know any fact about a universe (ie, you have a simulated universe running on a computer and whilst you don't have absolute knowledge of that universe if the software allows you have the ability to potentially interrogate it to know anything knoweable) could potentially be considered omniscience.
The interesting thing with that discussion, which relates to here, was that we are considering the idea of omniscience towards a realm under governance whilst leaving open to possibility of falling far short of such a standard in another realm.

So here is my response:
Bear in mind the example I gave in the original video. The cartesian-demon like example of God's God.  The idea that one such unknown unknown could be that God himself is a created agent, made by an even higher level entity in such a way that God's God never 'reveals' Himself to God; leaves no clues; fashions God's realm and his psyche such that He earnestly believes Himself to be the ultimate source. However, God is labouring under an illusion - an illusion from which there is no escape and no way of detecting or falsifying. At least, no way unless God's God decides to make things otherwise.

So this leads me to throw back two responses:
1)  First off, whilst the 'know everything that can be known' is in some ways entirely reasonable, and I take it deadly seriously, we can get into the same troublesome waters of those who claim it legitimate to limit God's omnipotence to just those things which are logically possible given his other attributes*. So I would suggest that such a proviso is reasonable so long as we do not then start to entertain allowing for factors that would make the total knowledge base sub-maximal. 
As a result I would point out that, as one such unknown unknown could be the existence of something as manifestly non-trivial as a God's God, having knowledge of the contents of this set (ie reassurances that certain things could not be contained in it) are critical ideas and ones which are perhaps too much to take in with regard to concessions we could make to omniscience in the traditional Abrahamic conception of the term. Further, clearly the existence of God's God whilst unknoweable to God (in the way I have proposed it) is most definitely not unknoweable. At the very least God's God knows of its own existence. So just as it would be deemed unacceptable for me to claim omniscience when I cannot absolutely declare the existence or nonexistence of God, how can we attribute omniscience to God when he is in the same epistemic position?
2) Why the traditional Abrahamic conception? This is where the rubber meets the road. Even if we were to concede a definition of omniscience that allows for unknown unknowns the concept of omniscience is not attributed to God just for lulz and I contend that such a definition - a non-absolute definition- is simply not fit for purpose.
The basis of God's Word as objectively defining moral law is critically dependent on God's omniscience, both in formulating His divine decrees (if you don't know the outcome of everything you canot ascertain that which is ultimately maximally good) and in terms of Him feeling justified in His own status of moral arbiter. However, the existence of a realm beyond God, or a higher power still, unbeknownst to God, smashes to pieces the foundations upon which this moral objectivity is claimed (I say claimed because as it stands I still hold it to be insufficient and baseless): God's morality becomes a subjective and arbitrary manifestation of however he was created, and however His realm was created, by the even higher power.

So that would be my defence:
i) It isn't just about the basic status of the set but the potential nature of the contents and the ramification of both such potentialities existing AND not being able to rule out their existence
ii) Such non-trivial potential contents makes uncertainty of their existence major epistemic issues for anyone claiming omniscience, not simply philosophical curios.
iii) Moral objectivity, as claimed by Abrahamic faiths, depends not just on omniscience but on a level of definition of omniscience that such a looser definition, as is proposed, clearly falls short of. Omniscience is proposed for a reason - if it does not do the 'work' claimed then that is what ought to concern us most, rather than a broader consideration of just what could potentially be cut from the definition in a broader sense of the word.

Thanks for reading :)

* I made a video on this
Defining Omnipotence: The Omnibenevolence Problem
I contend that such a restriction involves both reasonable and unreasonable claims. It is manifestly unreasonable to expect God to do things which are logically impossible. The fighting ground is over those things which are logically possible but not necessarily logically possible for God. In my view we need to differentiate between the acceptable (God cannot be expected to be me, even though it is logically possible to be me, as I have ably demonstrated) and the unacceptable (cannot do anything incompatible with omnibenevolence - which effectively means that in almost every circumstance his 'omnipotence' would allow Him one choice of action out of the myriad possibilities - effectively barely indistinguishable from impotence). My suggestion is that we define omnipotence not as everything logically possible for that agent but rather everything logically possible for a minimally constrained agent. In conclusion: if you wish to constrain God through omnibenevolence then He is no longer a minimally constrained agent and cannot be omnipotent by any reasonable measure.