karpierz

It's a 18 year long personal project where you get to build/train your own embodied AI, and then let it loose to wreak havoc on the world.

People are worked to death and the country has been in stagnation for 30 years. Time that could be spent dating is spent working and future prospects seem bleak.

And so someone did: https://old.reddit.com/r/MachineLearning/comments/k77sxz/d_t...

Google is free to publish whatever information they'd like to vindicate themselves. The fact that they don't is information.

Google is also free to sue for libel/defamation, but for some reason choose not to. This is also information.

You can say "well, we don't have the full picture, so we can't draw conclusions", but this will lead to never draw conclusions about the behaviour of anything, as you'll never have a full picture.

I wouldn't describe these as challenging, the prompts are fairly simple if you're capable of understanding pretty basic concepts.

> A pear cut into seven pieces arranged in a ring.

> A couple in formal evening wear going home get caught in a heavy downpour with no umbrellas

The work successfully demonstrates that the API lacks basic conceptual understanding by example.

> Pension funds inherently depend on population growth to avoid shortfalls.

Pension funds inherently depend on increases in production. Population growth is one factor, but technological development can also increase production.

> The other near-ponzi scheme is the real estate market. In my city of Toronto, average real estate prices have gone from 2x average income in 1972 to 16x average income today.

Hello from a Vancouverite! I broadly agree with this point, and I'm not sure how to work our way out of a housing bubble beyond popping it and dealing with the aftermath. Too many people are invested in the status quo, so any politician who tries to pop it will be crucified for destroying the savings of a large portion of the population.

> The suggestion in the report that science has been used to promote the dominance of Eurocentric views is a particularly egregious example of this misunderstanding.

https://en.wikipedia.org/wiki/Scientific_racism#Ideological_...

> There is just Science: a process that allows us to understanding how the world works by careful empirical investigation supplemented by rigorous theoretical analysis and predictions that are then subject to additional empirical testing.

I take issue with this definition. It uses a lot of words which seem meaningful, but are just arbitrary thresholds: "careful", "rigorous", "testing".

How do I know if I'm doing "careful empirical investigation" or just "empirical investigation"? Am I being rigourous enough to be "Science"? How do I tell if my "testing" is valid?

Is it "Science" if I think parachutes work? There's evidence, but it could simply be a placebo.

Yes, but unironically.

Sure, I'm not trying to argue about how to appropriately price the rate; just that it's variable.

I doubt their model is that people are perfect. I think their model assumes that if they set strong enough incentives, the human-error will only be a second order effect and can be safely ignored. I don't know how well they've estimated the political price for some of these incentives, and whether they have the political capital to expend to maintain this lockdown for much longer.

That is true, but the pertinent question is: how much less?

If the world is ending tomorrow, then much less. If we discover immortality tomorrow, then much more.

If you define socioeconomic status narrowly (gender, race, age, net wealth), then I was wrong to say that the success of European countries is rooted in their socioeconomic status. I'm happy to take that point back, and focus on why I don't think controlling for socioeconomic status is the right analysis.

However, even if the US shows better outcomes when controlling for socioeconomic status, the distribution of people within socioeconomic status is relevant.

If the US has 1000 impoverished people and 10 middle class people, and Germany has 10 impoverished people and 1000 middle class people, the relevant comparison is the German middle class to the US impoverished class.

In other words, you'd want to compare outcomes based on percentiles regardless of socioeconomic status.

This article needs Wittgenstein to edit it. The empirical claim is:

"Science" "blocks" "new ideas"

I add quotes because these words aren't well defined as is. The article presents the following as evidence:

When a researcher dies, another takes their places and starts a different stream of research. The issue with this as evidence is that it takes a very generous definition of "blocks". If I get hired for a job instead of you, have I "blocked" you from taking the job? It's also trying to imply that the now-dead researcher was stifling innovation, but fails to show that the new line of work is more novel than the existing work that the deceased researcher was working on.

The next piece of evidence is that work that cites unique pairs of journals will be more cited outside of the journal its published than within. But this stretches the definition of "new ideas". Interdisciplinary work isn't necessarily novel. Often, it will simply involve the application of known techniques from one field to the domain of another.

The third is an experiment where they condition on the novelty of papers according to keywords, and measure the likelihood of getting a grant. This also stretches the definition of "new ideas" into a new dimension, where unfocused work is considered more novel. They also claim that it's concerning that having cited the authors work previously increases the likelihood of the author getting a grant; as though knowing that someone produces quality work shouldn't improve the odds of funding for more work.

In summary, the issue with writing like this is it never bothers to actually define the thing its measuring ("novelty"), and instead borrows multiple other definitions ("novelty is interdisciplinary", "novelty doesn't cite popular works in the field", "novelty means having more keywords"). In doing so, it says nothing about novelty itself (instead talking about "novelty") and so doesn't provide any insight as to whether there is sufficient novelty within academic science.

Edit: A more concise way to make this point - https://xkcd.com/2610/

To be clear, max-flow/min-cost flow is not NP-Complete.

It's very compelling when it lines up with my worldview in a way that makes me feel validated. Then I can absorb it uncritically and regurgitate the conclusions to people without thinking about the evidence that lead me to the conclusions.

Unfortunately, I'm not a WASP so your 'horse theory' doesn't make me feel warm and fuzzy inside.

What does it mean to be "successful, controlling for socioeconomic status"?

Their success is defined by their peoples' socioeconomic status.

The article cites an anecdote about a young father who sat around the house arguing, fighting and drinking. He ends up getting kicked out of the house.

The 1960s version of this story instead ends with the husband beating his wife behind closed doors without repercussions.

Which is preferable?

> Better-educated people commit crimes at roughly the same rate, they are just far less likely to be caught.

While controlling for income? Or without considering confounders?

That's a guess, where's the validation?

What do you mean by "open sourcing" the Twitter algorithm?

What are the actual requirements that you're looking to satisfy by doing this?

This confused me at first too. The issue is that "X" is your performance, and "Y" is your perceived performance.

Say that everyone is equally okay at assessing themselves, and get within 0.1 of their actual performance (rated from 0 to 1). Then X and Y are going to be very correlated, as X - 0.1 < Y < X + 0.1. But X-Y will look like a random plot, since Y is randomly sampled around X.

The only case where X and Y wouldn't correlate at all is if people have no ability to assess their performance (IE, Y isn't sampled around X, but is instead sampled from a fixed range).

I think you're right on the formalism, and I shouldn't have brought that in. I've been out of grad school a while and should clearly brush up on my theory.

My main line of reasoning is:

1. If I say "this algorithm time-complexity is O(X)", I mean that all possible cases are in O(X).

2. If all possible cases are in O(X), then the worst case is in O(X).

3. If the worst case is in O(X), then any other case will also be in O(X). This follows from our definition of "worst case".

4. Therefore, "all possible cases are in O(X)" <=> "worst case is in O(X)"

5. Therefore, "this algorithm time-complexity is O(X)" <=> "this algorithm time-complexity is worst-case O(X)"

If your thought is that 1) isn't a formal definition, fair enough. It may have been a convention we adopted within the department. But aside from that, I think the reasoning is solid.

Other studies don't control for the amount being eaten. Time-restricted eating is a technique for eating less food, but aside from that it doesn't have intrinsic benefits.

I think it's all the union busting, anti-vax/anti-tax stuff.

Yeah, okay. You are correct here.

"Average-case of algorithm X" can be a function function (average number of operations for all possible inputs of length n). And asymptotic bounds can be applied to arbitrary functions from R -> R.

It is not redundant to say "average case O(n log(n))". It's not formally a correct usage of O(). But colloquially, it means that the average number of operations over all possible inputs on length n is X.

It is redundant to say "worst case O(n^2)". By definition, O(X) means that the maximum of the number of operations over all possible inputs of length n is at most X.

Theta(X) is not "average case" behaviour, it means "both Omega(X) and O(X)".

Ex: merge sort is Theta(n log(n)), but quick sort is not.

The author is likely saying "worst case O(n log(n))" as redundancy to emphasize that this is worst case behaviour.

True.

I was trying to say that hiding your face in a dating profile picture is an easy way to get ignored, but I went too far with it.

If you read the paper, you'll notice that the photos they use look nothing like dating profile pictures. Namely:

1. The faces are covered with a black box.

2. Standardized clothing.

3. The photos are straight shots (and for one dataset, side shots) taken in a blank room.

Dating profile pictures generally:

1. Prominently display your face.

2. Involve you wearing clothing indicative of how you present yourself.

3. Involve you in situations that are indicative of how you present yourself.

I hate this kind of research. Scrape some data, observe some correlations, and then imprint your view of the world onto it to draw some ad-hoc conclusions.

For example:

> In this study, researchers asked two different groups of women to look at photos of different men and rate how strong the men looked. Results showed that the rated strength of a male body accounts for 70 percent of the variance in attractiveness (this is a massive effect size).

Except that if you read the paper, you find:

1. "Furthermore, for two of the three regressions, the attractiveness measure became a negative predictor once strength ratings were controlled for. In other words, there appear to be cues in the male body that accurately indicate strength (and are detected as such by raters), but nonetheless are neutral or negatively valued when assessing attractiveness."

2. You know what else is correlated with strength? Mental health, income, social status, etc. You know what the study doesn't at all control for?

3. It turns out if you only give people a photo and have them judge attractiveness, they'll only look at the physique of the person. This doesn't mean it's how people view attractiveness in real-world situations where you have all sorts of other data.