Wednesday, July 18, 2018

Republican voters are completely insane

If you were hoping that Helsinki might be Donald Trump's Joseph-Welch moment, think again.  Donald Trump will not suffer any negative consequences from his disastrous and treasonous remarks.  This is why:
A new tracking poll from Reuters/ Ipsos on Tuesday showed that rank-and-file Republicans not only continue to support President Trump but refuse to believe he’s doing anything wrong. The most galling number from the survey is: 71 percent of Republicans approve of Trump’s handling of Russia... 
Overall, more than half of those polled (55 percent) disapproved of Trump’s handling of Russia and agree (59 percent) with U.S. intelligence findings about Russian attempts to tamper with the 2016 election. But, also worrying, is the fact that only 32 percent of Republicans believe that Russia attempted to intervene in the election.
This is really getting scary.  I used to think that Trump's claim that he could shoot someone in Times Square and get away with it was hyperbole.  I'm not so sure any more.


206 comments:

«Oldest   ‹Older   201 – 206 of 206
Don Geddis said...

And now, for a brief intermission: A parable about Cthulhu and ants: "This is the best explanation for higher powers I’ve ever really heard."

P.S. Thanks Peter, for offering the Jaynes reference.

Luke said...

@Peter Donis: (1/2)

> A good collection of data supporting this is in E. T. Jaynes's book, Probability Theory: The Logic Of Science. He uses Bayesian methods repeatedly to obtain results that are borne out by actual data, but which were either not reached at all by scientists in other fields, or were only reached after a much more laborious process of reasoning (that often ended up misleading people as well as being slower). Not all of his examples are in the social sciences, either; IIRC at least one is from thermodynamics.

Thanks for the Jaynes reference; if and when I need to do a deep dive into Bayesian inference, it looks quite promising as the kind of overview I like to start with. My favorite exposure to Bayesian inference so far comes from sensor fusion, when it was explained to me that Kalman filters are just Bayesian estimators. Before that, I just couldn't grok the seemingly random matrix multiplication.

> Luke: And if Bayesian inference is as awesome as you are currently portraying it, humanity is losing out tremendously every single second that Bayesian inference is not used to compare schools of thought in the social sciences.

> Don Geddis: … The sciences (especially social sciences and medicine) have been making decades (or centuries) of suboptimal reasoning (mistakes), because of their non-Bayesian thinking. …

> Luke: » Do you know this to be true based on anything empirical whatsoever, or have merely defined it to be true? « That is, can you show a correlation between optimality in reasoning in the social sciences and deployment of Bayesian thinking?

> Peter Donis: I've given one reference just now in response to Don.

As I hope the context makes clear, it is different to say that (i) "if the social sciences were to use more Bayesian inference, they would do better" and (ii) "it is quite reasonable to compare schools of thought in the social sciences with Bayesian inference". My contention is that (i) ⇏ (ii). Jaynes demonstrates (i), but not (ii). I never meant to contest (i); the reason I asked for explicit examples of it was to investigate the implicit induction from (i) → (ii). What I am imagining is that "The more Bayesian inference, the better!" actually stops being true well short of being able to compare entire schools of thought. And I'm more interested in what we are currently capable of doing, not what we might be able to do 200 years in the future. It is profoundly unfair for my current practice to be judged by promissory notes. (I suspect that "comparing schools of thought" as "comparing religions" via Bayesian inference, if possible, would require a massively collaborative effort.)

Luke said...

@Peter Donis: (2/2)

> If one person is trying to convince another person of the truth, there is no such thing as "consent".

Within a given formal system, I agree. Outside of everyone choosing to work within a given formal system, there are a plurality of ways to systematically represent phenomena. (Even measures of simplicity require some reference description language.) Unless you just meant acknowledging something like "there is a coffee cup next to my computer".

> If we are talking about interactions between humans, then your instinctive reaction against what seems to be "domination" in such an interaction is probably well-founded, because humans don't have complete knowledge of the truth and we're not very reliable at communicating it.

I think that human freedom exists, such that where we are going is a function of what we decide and not just what the laws of physics + initial state + intelligence-free randomness determine. You can force your preferences on me or we can find ways to cooperate. And yet, there is no such distinction with finite state automata. Everything is merely input; there is no "good" vs. "bad" input.

> However, Ron was talking about God, who, by hypotheses, does not suffer from those deficiencies.

If we are finite state machines, God surely would know the right pick-up lines to input into us. If instead there needs to be freely willed cooperation, then plausibly God need have no deficiencies for the status quo to obtain. What I suggest to those who think that we are FSMs and that there is nothing like what I've called "human freedom" above, ought to live as if that were true in every aspect of their lives. I'm not convinced very many do; I think most profess one thing while acting out another. I say that means we don't actually know what the professed thing looks like in a society. Possibly, the professed thing makes no sense whatsoever.

Luke said...

@Ron:

> > Then how are you using the Bernstein–von Mises theorem?

> To show that the initial choice of priors don't matter in the long run.

I've never seen a theorem applied via "a matter of taste". Either the theorem applies to the situation and you can show how via some mechanical procedure, or you have no idea whatsoever if your stopping point lies "in the long run".

> But if you want to demonstrate that God is omniscient, omnipotent and omnibenevolent and died on the cross to save us from our sins, that will be considerably more difficult.

The only thing that could be demonstrated to finite beings (us) is that God's a lot smarter / more powerful / more good than we are. As to Jesus' sacrifice, we would need an understanding of normative causation and as far as I know, we have nothing of the kind. It's all 'subjective', as if Cartesian dualism were actually true.

That being said, René Girard has done some pretty interesting things showing how scapegoating is foundational to culture and that if it weren't for the Jewish and Christian scriptures, we might never have come to see scapegoats as innocent. If people need to feel righteous and will not take full responsibility for their [in]actions, what happens to the residue? If the alternative to my faith is to punt on this issue, that is a distinct loss.

> I never said they couldn't possibly be super-natural. I simply don't see any evidence that they are anything but a human invention.

When you remain consistently silent on what would qualify them as super-natural, it is easy to suspect that per your means of evaluating the evidence, nothing would qualify. And just to be clear: 'super-natural' in this context reduces to "that which exceeds human capability". My issue here is that unlike F = GmM/r^2, which can be falsified by something almost indistinguishable (say, ^2.001), I suspect the only thing that would falsify your current notion of 'natural' is something really crazy. If God actually wants to help us explore reality and be better to each other, doing something really crazy is profoundly undesirable. So what you seem to have done—emphasis on "seem"—is render this particular understanding of God undetectable until the shit really hits the fan and all human efforts to fix it fails.

> This isn't on you. It's on God.

Actually, I'm saying that you've left yourself open to every rationalization in the book when the 10-year mark has arrived, for saying that whatever it was it wasn't God.

Ron said...

@Luke:

> I've never seen a theorem applied via "a matter of taste".

Are you being intentionally obtuse? The theorem itself is not a matter of taste. The theorem tells you that given the same evidence, two Bayesian reasoners will converge to the same posterior probability regardless of their priors. What is a matter of taste is choosing the cutoff threshold where you say, "That's good enough, P(X) is close enough to 1 that I'm going to proceed as if it were actually 1."

> scapegoating is foundational to culture and that if it weren't for the Jewish and Christian scriptures, we might never have come to see scapegoats as innocent

That assessment shows a profound ignorance of history. Christians were scapegoating well into the 1700s. Scapegoating didn't end with the advent of Christianity, it ended with the enlightenment.

> If we are finite state machines, God surely would know the right pick-up lines to input into us.

We are in fact finite-state machines. You even concede this yourself:

> The only thing that could be demonstrated to finite beings (us)

Luke said...

@Ron:

> Are you being intentionally obtuse?

No. The fact that you're currently within some acceptable ε of someone else after some number of updates doesn't mean that you both won't continue changing well outside of ± ε. Now, perhaps you were thinking of converging on only 0 or 1? If that's the case, I want to know how to think of all the evidence which strengthened the belief that reality is Newtonian in the 1800s; did that make it harder to understand quantum physics? (I'm trying to get at just what beliefs are approaching 0 or 1.)

> > scapegoating is foundational to culture and that if it weren't for the Jewish and Christian scriptures, we might never have come to see scapegoats as innocent

> That assessment shows a profound ignorance of history. Christians were scapegoating well into the 1700s. Scapegoating didn't end with the advent of Christianity, it ended with the enlightenment.

I didn't say that scapegoating ended with the advent of Christianity.

> We are in fact finite-state machines. You even concede this yourself:
>
> > The only thing that could be demonstrated to finite beings (us)

I don't see how "finite" ⇒ "finite state machine".

«Oldest ‹Older   201 – 206 of 206   Newer› Newest»