At least this tells me I didn't make a silly mistake in my post. Thank you for the feedback.

As for your objections,

All models are wrong, some models are useful.

exactly captures my conceit. Reductionism is correct in the sense that is, in some sense, closer to reality than anti- or contra-reductionism. Likely in a similar sense that machine code is closer to the reality of a physical computation than a .cpp file, though the analogy isn't exact, for reasons that should become clear.

I'm typing this on a laptop, which is a intricate amalgam of various kinds of atoms. Hypothetically, you could explain the positioning of the atoms in terms of dense quantum mechanical computations (or a more accurate physical theory, which would exist ex hypothesi), and/or we could explain it in terms of economics, computer science and the vagaries of my life. The former strictly contains more information than the latter, and subsumes the latter to the extend that it represents reality and contradicts it to the extend it's misleading.

At an objective level, then, the strictly reductionist theory wins on merit.

Reductionism functions neatly to explain reality-in-general, and even to explain certain orderly systems that submit to a reductionist analysis. If you want completeness, reductionism will give you completeness, at the limit. But sometimes, a simple explanation is nice. It'd be convenient to compress, to explain evolution in abstract terms.

The compression will be lossy, because we don't actually have access to reality's dataset. But lossy data is okay, and more okay to more casual the ends. Pop science books are very lossy, and are sufficient for delivering a certain type of entertainment. A full reprinting of a paper's collected data is about as lossless as we tend to get.

A lossless explanation is reductionist, and centribus paribus, we ought to go with the reductionist explanation. Given a choice between a less lossy, very complex explanation and a lossy, but simple explanation, you should probably go gather more data. But failing that, you should go with one that suits your purposes. A job where every significant bit digit of accuracy matters chooses the first, as an example.

I think it's not so much a sum of properties as a union of property sets. If a system has a property that's not a part of a union then it's "more than the sum of its components". On the other hand I find the notion of something being "more than the sum of its parts" about as annoying as the frequent ads with "1 + 1 = 3 Buy two and get one for free!" equation. That is, very annoying.

I know of an old prime number that happens to end with a 2.

It seems interesting that a lot of spiritual experiences are something that happens in non-normal situations. To get them people may try denying food or sleep, stay in the same place for a long time without motion, working themselves to exhaustion, eating poisons, going to a place of different atmospheric pressure or do something else they don't normally try to do. The whole process is suspiciously similar to program testing, when you try the program in some situations its creator (evolution in case of humans) haven't "thought" much about. And then sometimes there are bugs. And if you don't follow the protocols for already discovered bugs you either risk crashing something really important or getting nothing at all. Bugs are real and may give a valuable information on the program's inner workings, but they're not "the final truth about the underlaying reality".

The belief of the revelatory nature of spiritual experiences may be a result of a "just world" bias. When you get your reward you've been working for for years, it's easier to believe you understood something profound about the reality rather than that you've discovered an error in your brain. If that's the case then "If you spin a lot, you'll get vertigo" or "if you sit on your hand long enough, there would be strange feeling there" or "look through the autostereogram picture to see it in 3D" may be thought of as spiritual experiences, but they're too easy and mundane for that.

I've had similar thoughts in the past few days. It does seem that utilitarianism merely prescribes the moral action, without saying anything about the goodness or badness of people. Of course, I've seen self-identifying utilitarians talk about culpability, but they seem to be quickly tacking this on without thinking about it.

Isn't the question of someone being a good or a bad person at all a part of virtue ethics? That is, for a utilitarian the results of the bystander's and murderer's actions were the same, and therefore actions were as bad as each other, but that doesn't mean a bystander is as bad as the murderer, because that's not a part of utilitarian framework at all. Should we implement the policy of blaming or punishing them the same way? That's a question for utilitarianism. And the answer is probably "no".

I'm asking if there is a rational non-meta reason to believe they do "stop at the neck" even if we throw away all the IQ/nations data.

Of course there are. The standard argument is that the history of human evolution suggests that increased intelligence and favorable personality traits were strongly selected for, and traits which are strongly selected tend to reach fixation rather quickly.

People in our society differ in how they think about genetic differences. There are people who think that race matters a great deal and other you think it doesn't matter. It's useful to have a metric that distinguishes those people.

If you have that metric you can ask interesting questions such as whether people who are well calibrated are more likely to score high on that metric. It's interesting whether the metric changes from year to year.

That means the question tries to point at a property that people disagree about. In this case it's whether genetic differences are important. The question doesn't define "important" but there are various right wing people such as neoreocons and red-pill-folks who identify with the term "human biodiversity". The question doesn't try to ask for a specific well-defined belief but points to that cluster of beliefs. It's the same way that the feminism question doesn't point to a well-defined belief. You don't need a well-defined belief to get valuable information from a poll.

The question made it into the the survey because I complained about the usage of tribal labels such as liberal/conversative where people have to pick one choice as a way to measure political beliefs. I argued that focusing on agreement on issues is more meaningful and provides better data.

I don't think "ignoring the context" is well described as going deep. Part of critical reading is to think about why someone writes what they write instead just trying to focus on the literal meaning of words. It's rather only engaging with the surface.

I am not sure what are you arguing. The fact that there are important genetic differences between populations at the medical level is uncontroversial. The controversial issue is whether these differences, as some people put it, "stop at the neck".

the question singles out intelligence and personality traits for no apparent reasons but controversy

Nope, there are apparent reasons. Intelligence of the populations is massively, hugely important, much more so than lactose intolerance or propensity for exotic diseases. See e.g. this or this.