alskdfji

This was the most depressing part to me: "Another person I know is in a similar situation because his group won’t talk to people who aren’t already employed."

It's a long story, but both my wife and I have most recently been at places, in a small-medium city, that have had huge turnover in the last several years due to mismanagement (her entire unit, except for her, left twice while she was working with them; we've lost about 30% of our employees in the last few years). She was pressured to resign after returning from her maternity leave because, as far as I can tell, they figured out they could cut her position and offload the work onto others while she was gone. So she's stayed home to raise our child. I haven't lost my position, but every week brings a new clusterfuck of horrors that I'm running out of ways of coping with.

The problem is, now she's wanting to return to work, and has a job offer, but it's in another state, and it's created a situation where basically I have to give up a job and probably career, or she has to run the risk of her career dying off because of lack of opportunities in the area.

So if I take time off to watch our child while my wife rebuilds her career, and maybe move to a different career myself, somehow I'm penalized? Pardon my language, but fuck that.

I'm also in my early 40s, so there's that.

I'm starting to feel like my life is ending, really, like opportunities are just vanishing left and right. It's odd because up until this point I always felt optimistic, like there was always something out there for us. My wife comes from an elite school, we both graduated with Ph.D.s from a program in the top 5 in our field (according to dubious rankings), if you're into that. I don't mean this narcissistically, but I just feel like there's this huge discrepancy between what I know my wife and I are, in terms of work ethic and competency, and what our opportunities are.

The human visual system is susceptible to misperception, sure. This has been known for awhile. I also would agree that sometimes single pixels could probably confuse people substantially.

However, if you look at the images in the article, I think it's telling about how much more sophisticated human perception is. The authors even have a note in the first image along the lines of "look carefully, because you might have trouble locating the aberrant pixel."

I think these types of attacks speak volumes about the fragility of DL optimizations: I think there's more overfitting going on than people acknowledge or realize. My sense from reading in working with some ML things is that this extends to natural language data as well.

DL systems are often highly optimized to a particular test set, and might do well on cross-validation to other exemplars of that test set, but that's not the same as generalizing across different types of test sets.

Maybe I'm wrong, though--I think DL NN models have been fantastic, but there's a certain amount of hype in ML in general.

I have mixed feelings about what you're saying and about the article.

On the one hand, I completely empathize with what you're saying--it's mystifying to me when I see practices continue when there's so much empirical data about them to suggest they shouldn't be used. In the US less-well validated procedures are on the wane, but they still appear in use from time to time.

On the other hand, although I agree Lilienfeld's paper is excellent, buried in it is the complicated truth that some projective tests do demonstrate evidence of validity. There's also the tricky issue that a test is not just a stimulus set, but a procedure, and even certain ways of the Rorschach have proven useful empirically. But all these things have been forgotten or ridiculed because of the current anti-projective zeitgeist. That is, certain projectives, used in a certain way, are invalid, and it happens those have been the most popular, but if you take another hard look at the data, it's not true of projective tests in general.

The article provides a nice overview of the Rorschach, but the end is puzzling to me. The MMPI is one of the best-validated behavioral measures in history, and the examples are so misleading that I wonder about the author and Annie Paul's motives. An item such as ‘My mother is (or was) a good woman’ isn't interpreted in isolation, and even if it was interpreted in such a way, it has predictive value in the same way as your social media history that so many services are starting to scrape for prediction also. It's an interesting article that ends on a note of incompetent ignorance.