The opposite is the case, of course. Evidence is always evaluated against expectations. If your fingerprints are on the crime scene, that may make you a suspect. But if you’re the butler, and the victim was killed in the study you regularly clean, that same evidence is just what you’d expect. However, if your fingerprints are on the murder weapon as well, you’re under suspicion again. Except if it was actually the knife you use daily to chop vegetables. And so on: just abstractly pointing to some given evidence doesn’t suffice to make a determination. Hence, you also need to posit what evidence—what evil—to expect (or not to expect) given the existence of a tri-omni God in order to make a convincing case.
Simple; none. Such a god could make “evil” as impossible as accelerating past the speed of light is in real life.
Which, as explained before, would take us back to the logical PoE that the rest of us have left behind about 600 posts ago. (Actually, even the logical PoE usually features an argument rather than just a bald assertion.)
neverm
This isn’t the opposite, it’s highlighting my point.
The evidence counts towards a claim. “Well, there might be reasons” doesn’t allow us to handwave that evidence, or else we could always handwave evidence. Only an evident mitigating factor gives us a context in which to re-evaluate evidence.
Which we don’t have in e.g. the “psychopath god” example that I’ve given.
No, it takes us back to the issue that the meaning of the words “omnipotent” and “omniscient” keep being carefully ignored, because paying attention to them undercuts all the nonsensical handwringing about “Ooooh, but what about necessary evil?!”
Degrading the power of God to less than that of humans in an attempt to evade the “problem of evil” doesn’t make it go away. It just means that you’ve invented a really pathetic version of God.
But nobody is saying ‘well, there might be reasons’. The point is that just there being fingerprints alone doesn’t allow us to make an evidential decision; the question is whether we should expect there to be fingerprints. If you don’t happen to be the butler, then we can say that there probably should not be any of your fingerprints around, so the fact that there are is surprising, and makes you suspicious.
But in the example of God, you’re pointing at just abstract evil as evidence. But we know that some evil is consistent with the existence of God, due to the failure of the logical argument. Hence, abstract, unspecific evil is not evidence against that particular hypothesis, in the same way that just fingerprints alone, without an expectation that there shouldn’t be any, isn’t evidence in the murder case. So any cogent formulation of the evidential argument starts exactly there: points to some evil that we probably should not expect if there is a tri-omni God—which is exactly the same as pointing to the fact that we should probably not expect your fingerprints at the scene, given that there’s no reason for you to be there.
It doesn’t follow. We know that there was, is, or will be at least some incriminating evidence in the world. We may have an independent basis to think it may be found in a given case–the foundation or providence of the evidence.
Same for the black swan. We have an independent reason to believe there actually is at least one black swan in the population from which we draw our sample. This knowledge is absolutely necessary for us to rationally believe the probability of a given swan call belonging to a black swan is nonzero.
Sneaky Godzilla is disanalogous because we have no independent basis to believe the creature actually exists (as opposed to being logically coherent).
~Max
No we don’t. I mean, it’s irrelevant anyway, but hypothetically there may be no crimes committed today, or ever again. We don’t require making claims about the set of crimes to process evidence.
I think you’re trying to think of any difference you can between the situations, because ultimately your argument (like all defenses or proofs of God, it seems), is special pleading.
If this is an allusion to “set of universes” reasoning again, then the population also contains infinite Godzillas, so we can reason that this universe has some Godzillas, by this reasoning.
If you’re only talking about this universe then no, we don’t know that the population contains even one black swan. The possibility of them existing is worthless; the tooth fairy has a possibility of existing.
I’m curious though what was the gotcha (as you put it) if I had accepted these premises.
(missed edit window)
I missed that you said “was, is, or will be”. So yes, it’s true that we know that there has been incriminating evidence in the past. It doesn’t tell us we’re going to find any in the future, or related to any specific crime.
So my point stands: making claims about the set of all evidence is not how we process evidence; it rests on its own merits.
It’s not that hard. Fingerprints at the scene, their mere existence, on their own/in the abstract tell us nothing. If we have grounds to believe that in this concrete instance, there should not be such fingerprints, e.g. that the person to whom they belong isn’t expected to have been at the scene, then the same evidence casts doubt on the hypothesis of that person’s innocence. Whether evidence counts against a particular hypothesis depends on whether that evidence, given the hypothesis, is surprising.
In the case of God and suffering, from the failure of the logical PoE, we know that suffering is possible even if God is all-knowing, all-powerful, and all-good. Thus, given the hypothesis of such a God’s existence, the mere fact of suffering is not surprising. If something that’s possible happens, you shouldn’t be surprised. For it to be surprising, we would need additional information, like, say, that suffering is very, very unlikely, and we should observe very little of it, in some sense. It’s hard to see how to substantiate such a premise, but if you were to do it, then—and only then—should we be surprised to observe more suffering than that.
That’s why a better tack is to concentrate on specific instances of evil, taking a concrete case as a starting point—which, once again, is what every formulation of the evidential PoE actually does. Your fingerprints probably shouldn’t be on the murder weapon if you’re innocent, yet, they are: that’s surprising, hence, casts doubt on your innocence. Animals probably shouldn’t burn to death in agony alone in the forest if there’s a tri-omni God, yet, they do: that’s surprising, hence, casts doubt on the existence of such a God.
This is nothing but the perfectly ordinary way to use evidence to adjust one’s beliefs. Evidence in a vacuum isn’t evidence; only once we have grounds on whether it is in line with a given hypothesis can we appeal to it to adjust belief in a rational fashion.
You are skipping the first step which categorizes an observation as evidence supporting a hypothesis in the first place. As I wrote above, we only reach the evidential problem of evil after defeating the (much simpler) logical problem of evil. In order to do that, the hypothesis must be modified. No longer does tri-omni God want zero suffering; now tri-omni God only wants zero gratuitous suffering.
You brought up criminal process. We convict people of crimes only after determining, beyond any reasonable doubt, that the accused is guilty. Our hypothesis predicts that the accused is guilty, and each piece of incriminating testimony increases our confidence in the prediction. As we hear testimony, we must decide whether or not the testimony is incriminating in the first instance. Only incriminating testimony increases confidence in our prediction. Not all testimony is incriminating; some is neutral or even exculpatory.
If the first witness testifies that he is a detective who found an unmarked gun at the scene of the crime, that testimony is neutral. The detective’s testimony neither increases nor decreases confidence in our prediction, at least not until we re-evaluate this evidence in light of later testimony (such as a witness who saw the accused carrying a similar gun). This is evidence, yes; but it is not incriminating.
If you hear a witness testify that the accused was in another place at the time of the crime, that testimony is not evidence consistent with our hypothesis (guilty). The exculpatory testimony counts toward a claim, yes, but the alternative hypothesis (not guilty).
It is not. There are no known sneaky Godzillas. Similarly, there are no known tooth fairies, unless you are a child who believes in the tooth fairy and has the coins to prove it.
Black swans are native to Australia, and for whatever reason there was a flock in Lakeland, Florida, USA. Regardless, the predicate term was conditional:
If I know some swans are actually black, but all swans sound alike, then every time I hear a swan I have some evidence that it is black. Most of the time, that evidence is mitigated by sight of white feathers. But in the first instance, hearing a swan cry provides at least some minimal evidence that it is a black swan… If I listen to enough swans, all else equal, there is an increasing probability that I will come across a black one.
~Max
That would be absolute madness. We’d be saying that we can’t doubt that omnimax God exists until we first prove that he doesn’t. I can’t suspect the butler did it until I first prove that he did.
Once again, we’re engaging in special, one-off reasoning rules to try to make this defence of the problem of evil work, rather than how we reason about anything else.
You’ve written a lot there about how the criminal process works, that is largely trivial (I could dispute that we begin with predicting that someone is guilty, but it doesn’t matter). But you haven’t addressed the point, which is that it shows (one of) the problems with the gratuitous suffering categorization.
We don’t need to say in advance what data will be incriminating. We don’t need to set a threshold. And we certainly don’t need to rule out that an innocent explanation is possible, because that is almost always hypothetically possible.
By that standard, all evidence would need to be thrown out – e.g. we can’t say that finding the murder weapon in a person’s bedroom is incriminating because it’s possible an innocent explanation could be found.
That’s not how suspicion, and circumstantial evidence works.
And all I am saying is we do the same with the problem of evil. So, no threshold in advance. No need to say upfront if a particular data point is gratuitous or not. We make a determination based on all the data we have. We have countless murder weapons in God’s bedroom. And, based only on what we know right now, “omnimax god exists” is a distant third behind “god doesn’t exist” and “god exists, but isn’t omnimax”.
And there’s no known (universal) necessary suffering.
I’m aware of that, but I’m using the term in the same way that you were, as something we don’t know exists. The possibility of a black swan existing in some universe does nothing to establish that any do in this one.
No. The logic is: if the logical PoE was sound, then any evil at all deductively proves God’s non-existence. But it isn’t sound. Hence, we can’t deductively prove God’s non-existence. Consequently, we need to argue inductively against the hypothesis of God’s existence. The evidential PoE is the best ‘leftover’ avenue of attack after the logical PoE has been defeated.
This of course immediately also establishes that not just any evil at all serves as evidence against God’s existence. From the failure of the logical PoE, we know that evil is possible even if God exists. Hence, it’s not surprising to observe evil, in the abstract, if God exists. That what is possible actually happens is no surprise—if it’s possible that it rains tomorrow, you shouldn’t be surprised if it does.
Things would be different if we had grounds to believe that it should be unlikely for there to be evil if God exists. Beyond the possibility of evil given God’s existence, however, the failure of the logical PoE tells us nothing. Large and small amounts of evil are equally consistent with this possibility. Hence, once again, we need to introduce additional information.
Mere observations aren’t evidence. Observations only become evidence against a hypothesis if the hypothesis entails that you shouldn’t expect the evidence—if, given the truth of the hypothesis, the observation is surprising. The mere existence of evil, because of the failure of the logical PoE, is not surprising. If you just observe fingerprints at the scene, that doesn’t tell you that the person whose fingerprints those are is suspicious. Only if there is no reason for them to be at the scene—which will be the case for the great majority of people, but still—does this qualify as evidence against their innocence.
Or to put this more succinctly: because you can’t prove the butler did it, you have to make an evidential case against his innocence.
If your position is that the concept of a “butler” is logically incompatible with a dead master, why would you bother listening to someone who says evidence increases confidence that the butler killed his master? Either the the master is alive or there is no butler. It’s a waste of time to speak of circumstantial evidence until you sort out your definition of “butler”.
The logical problem of evil says God is logically incompatible with any evidence of evil or suffering, full stop. It is an a priori paradox. If you want to argue that observed suffering increases confidence that God doesn’t exist, without first disarming the logical problem of evil, you are spilling ink. That is why, when you speak of an evidentiary problem of evil, I assume the premises necessary to disarm the logical problem of evil; to wit: only gratuitous suffering is evidence against the existence of tri-omni God.
As for your second point, raw data without any reference points is meaningless. You may not need to say in advance what data will be incriminating, but it is impossible to judge evidence without prior knowledge to serve as a reference. You must set some kind of threshold by which you can determine incriminating versus not incriminating (mutually exclusive alternatives), or incriminating versus neutral versus exculpatory (two alternatives and a null). Your determination does not have to be perfect, binary, or even accurate, but you must make an initial determination and that determination must be based on prior knowledge. The Sun has risen every day for 4.5 billion years, but that does not give us 1.6 trillion data points supporting the inference that the butler killed his master. It gives us 1.6 trillion data points that have no known import. You have to have knowledge going in to evaluate evidence in the first instance.
If you have moved past the logical problem of evil, that knowledge is that God is incompatible with gratuitous suffering. The more evidence you have of gratuitous suffering, the more likely it is that tri-omni God does not exist. But until you add another premise–which you have consistently argued is unnecessary, to my dismay–no amount of evidence of mere suffering, unqualified, supports the inference of tri-omni God’s nonexistence.
~Max
I think this point needs to be pressed and amplified (unfortunately, @Mijin has seemingly decided to either ignore my posts or myself completely, so I’m not sure what good this will do).
Fingerprints on a crime scene on their own are not evidence of guilt. We know it’s possible for fingerprints of an innocent person to be at the scene. Fingerprints you wouldn’t expect to be there—gratuitous fingerprints—are such evidence: their presence makes the hypothesis of innocence less likely to be true.
Evil on its own is not evidence of God’s non-existence. We know it’s possible for there to be evil even if a tri-omni God exists (from the failure of the logical PoE). Evil that you wouldn’t expect to be there—gratuitous evil—is such evidence: its presence makes the hypothesis of a tri-omni God less likely to be true.
Just as you can’t point to the mere presence of fingerprints as incriminating evidence without further information, so you can’t point at the mere existence of evil as evidence disconfirming God’s existence. You have to point to evidence such that its presence would be unlikely or surprising if the hypothesis were true to lower credence in the hypothesis.
Again, this is absolutely backwards reasoning. As you say, the logical problem of evil makes the stronger claim. That’s why, no, in fact I can say there is insufficient grounds to say suffering and omnimax are formally incompatible but still say the weaker claim holds; that suffering gives us reason for doubt.
Just as I can say that there’s insufficient grounds to conclude that my wife is cheating on me, but also that there is reason to suspect that that’s the case.
Or indeed any empirical claim about anything.
I am simply reasoning in the same way we all do about any observation. This “proof first” reasoning is a special plead for a defence of God.
None of it is in advance. So this need to set thresholds in advance for evidence about suffering is again trying to make a special case.
And in terms of “prior knowledge”, I thought we’d dealt with this in my post 708. We only needed the definitions of the terms going in, we didn’t need to cross-reference across other cases and examples.
First, while the logical problem of evil makes a strong claim (God is always incompatible with suffering), it is defeated with weak claims (God is always compatible with necessary suffering; God is always incompatible with gratuitous suffering). Disarming the logical problem of evil leaves a residual hypothesis: if gratuitous suffering exists, tri-omni God does not.
Second, it is always impossible to evaluate evidence without using pre-existing knowledge to determine whether the evidence supports the hypothesis. This body of knowledge can change over time as evidence is added to it, the threshold need not remain fixed over the life of the experiment, and determinations can be provisional or probable rather than certain and final, but each determination is necessarily dependent on reference knowledge. Otherwise there is no way to separate noise from probative evidence.
~Max
None of these are claims that I am making though: you are talking about things that are part of the premise of (one of) the problem of evil and/or entailed by the definition of “omnibenevolent”.
Once again: you can take the position that the PoE is ill-formed because suffering itself is not inherently bad. Or you can take the position that suffering is inherently bad, but that the best possible world contains some suffering.
Or you can believe both.
What you can’t do, is jump around, such that when I’m explaining the issue with one of these defences, retreat back to saying I need to prove one of the things that’s a premise of the PoE. Something that the majority of people – theist or not – would accept, and that is that an omnibenevolent god wouldn’t cause gratuitous suffering.
Again, you’ve used this talking point over and over but as I’ve shown with examples like criminal cases or vegetarianism etc, we only need the definitions of the terms and then which hypothesis the circumstantial evidence points more towards.
I’ve done it line-by-line and like-for-like showing that if we are evaluating evidence in a consistent way, we can suspect that the butler did it, and that Bob is not a vegetarian and that God, if he exists, is not omnimax.