smu3l

Conceptually the same thing though. Claude.md is sourced at the beginning of the session, so will be pretty far back in context, just like user prompts from the beginning of the session.

Banned from what?

I think the order of enemy moves is deterministic. So if piece 1 blocks piece 2 from attacking you, piece 2 can attack if 1 moves out of the way.

Why those countries? Does google have weaker internal security for employees in those countries?

A couple of decades ago I was taking a class at UNC on survey sampling methods. Topics included designing sampling schemes that were efficient in the statistical and actual cost sense, developing variance estimators based on your sampling scheme, etc.

For example if you want to observe and measure some attribute of classes at public schools in your county, it might be infeasible to send data collectors to all of 15 schools, but the marginal cost of measuring additional classrooms at the same school once you're there is minimal. So, how many schools should you visit and how many class rooms per school given a budget and assumptions on inter and intra school variation?

We had had a group assignment to estimate the average circumference of trees on campus. Our initial plan something like 1) get a map of campus and split it into zones 2) sample zones randomly 3) everyone goes to a few (small) zones and tries to roughly map out the trees there 4) sample again from those trees and physically measure them. This would mean running around campus for at least a few days if we wanted to an honest job. And it was a rainy spring in North Carolina.

However, one of my group mates had a stroke of brilliance and decided to email the grounds department. To our surprise they were able to provide us with a full list of every known tree on campus as well as GIS data with locations. So we were able to do a legitimate simple random sample which was optimally efficient from in terms of both variance and time-in-rain.

In conclusion I'm pro list-of-individual-trees.

The last SE had a smaller screen but the actual device was larger than the last mini (13 mini). Both are discontinued now, sadly.

The 16e is the smallest phone in the current lineup, which is bigger than both.

I really like the 13 mini, I had been on Android since since maybe iphone 4. Not sure what I'll do when my current phone dies, I don't like having a big phone in my pocket.

Scala (and Java) has a typed Dataset api.[0] pyspark only provides the Dataframe API, which is not typed.

[0] https://spark.apache.org/docs/latest/sql-programming-guide.h...

One, if you mean the directory containing the css files (helpfully called 'css') or two, if you mean one of the files.

Never heard of ibis before but the front page give a pretty good overview of what it is to me at least. Looks like a dataframes api that can be executed on a variety of backend engines, both distributed and local.

Somewhat analogous to how the pandas api can be used in pyspark via pyspark pandas, but from the api -> implementation direction rather than (pandas) implementation -> api -> implementation maybe?

I don't think GP is suggesting the time was wasted. Rather the opposite.

The article references a tweet about an accelerator pedal problem. It doesn't specifically say what the issue is, but I think it's referring to the issue where the decorative plate on the top of the pedal can come off and cause the pedal to jam in the fully depressed position.

https://arstechnica.com/cars/2024/04/tesla-stops-cybertruck-...

> Researchers at universities also do these kinds of things because it helps them advance their careers.

This is a huge problem and in my opinion is mostly due to bad incentive structures and bad statistical/methodological education. I'm sure there are plenty of cases where there is intentional or at least known malpractice, but I would argue that most bad research is done in good faith.

When I was working on a PhD in biostatistics with a focus on causal inference among other things, I frequently helped out friends in other departments with data analysis. More often than not, people were working with sample sizes that are too small to provide enough power to answer their questions, or questions that simply could not be answered by their study design. (e.g. answering causal questions from observational data*).

In once instance, a friend in an environmental science program had data from an experiment she conducted where she failed to find evidence to support her primary hypothesis. It's nearly impossible to publish null results, and she didn't have funding to collect more data and had to get a paper out of it.

She wound up doing textbook p-hacking; testing a ton of post-hoc hypotheses on subsets of data. I tried to reel things back but I couldn't convince her to not continue because "that's how they do things" in her field. In reality she didn't really have a choice if she wanted to make progress towards her degree. She was a very smart person, and p-hacking is conceptually not hard to understand, but she was incentivized to not understand it or to not look at her research in that way.

* Research in causal inference is mostly about rigorously defining the (untestable) causal assumptions you must make and developing methods to answer causal questions from observational data. Even if an argument can be made that you can make those assumptions in a particular case, there is another layer of modeling assumptions you'll end up making depending on the method you're using. In my experience it's pretty rare that you can really have much confidence that your conclusions about a causal question if you can't run a real experiment.

What is the precise definition of static typing? Both the article and many comments in this thread refer to static typing or static type hints in Python. But as I've always understood it, static typing means that types are checked and enforced at compile time. And if python is not compiled, the notion of static typing in python does not make sense. So do I have it wrong? Or is the term ambiguous?

Multiarmed bandits and contextual bandits are essentially causal inference with a cooler name. You can formulate both with a potential outcomes/couterfactual framework, and contextual bandits typically is presented that way.

(Bandits are often presented as a loop where you control the policy collecting data and update it frequently, but that does not have to be the case.)

Recommender systems and search/counterfactual learning-to-rank can be thought of as an extension to counterfactual bandits as well.

I want to say one word to you. Just one word. Are you listening? ... Bandits.

Ah! It's only on that post, not the rest of the site. Clever.

Indeed. I just set scrollbars to always show on my Macbook per the author's suggestion and there are 4 scrollbars on their page!

Makes me wonder if I'm missing the point of the fairly simple post?

This is always true: you're always making some strong assumptions when trying to make a causal claim. Using a causal framework, DAGs in this case, at least makes those assumptions explicit.

Chronology isn't enough though. Roosters crow before dawn.

This is a special case of sampling bias.

What kind of work was it you do? :)

Oops I did not see your response until now.

I agree, changing the model changes the estimates, because the parameters you are estimating change.

However, given one misspecified model, the parameters of that model are still well defined, though they may not have the interpretation they would if the model was correctly specified. As OP called it, this is the "best fit line", and is a projection of the truth onto your model. E.g. for a simple linear regression of Y on X, where the true conditional mean of Y given X is not linear, there is still some "true" best line. This line depends also on the distribution of X, though it would not if the model was correct. Estimates from linear regression will converge to the parameters of this line, though using the usual standard errors will be wrong.

There's a very general theorem or corollary that covers this in Asymptotic Statistics by van der Vaart. I think in the chapter about M estimators, right around where MLEs are covered, but I don't have it in front of me.

From the semiparametric perspective, you can still make correct inferences about the estimated parameters even if the model is not correctly specified, as long as you use the so-called robust estimator of the variance.

This is not correct. Logistic regression can be used for classification, true, but it can also be viewed as a way of estimating the conditional mean of an outcome variable that has a Bernoulli, or binomial distribution, depending on the formulation.

There are many ways to evaluate all of these methods, and for classification you may favor something else, but it's completely reasonable to use the (cross validated, or not) empirical risk for both logistic and linear regression. That would be a negative log likelihood in both cases, from the Bernoulli/binomial distribution for logistic regression or the normal distribution for linear regression.

Not true, you do not have to control for everything. You have to control for everything that may have an effect on both the treatment and the outcome. That implies you have to assume that everything you don't control for does not affect both the treatment and outcome, (though something can affect one). This assumption is not testable but sometimes reasonable.

You also have to assume that it's possible for every person/unit of study to have a non-zero probability of receiving either treatment for all levels of the variables you are controlling for in order for the effect you're estimating to be defined. That is more likely to be violated the more things you have to control for.

Those are blue, these are black. They also have a different name. Finally, it appears that the content of at least some of the cards is entirely different between the two games. (I cannot say to what extent because I've only seen the sample of cards that are on the websites.)

There's a whole page[1] dedicated to defining "free software" in the same Philosophy section. Of course that page also defines the "copyleft" pun which is loaded and confusing on its own.

[1] https://www.gnu.org/philosophy/free-sw.html

The drug itself is very cheap. With an Rx you can purchase an ampule with about 3 doses for < $10.[1] You can administer it with an insulin syringe, which I don't think requires an Rx. Of course administration is a lot more tedious and requires some training.

I worked at a summer camp where we would carry 1 or 2 pens plus a bunch of ampules if we were in a wilderness situation where we couldn't get to emergency care and needed more than one dose.

[1] http://www.buyemp.com/product/epinephrine-ampule

Yes, this is incorrect.

Also the next sentence is wrong:

>You’ll often see statistical power conveyed as P90 or 90%. In other words, if there’s a 90% chance A is better than B, there’s a 10% chance B is better than A and you’ll actually get worse results.

Having a test w/ 90% power means that if A is truly better than B (for one sided) or A is truly different than B (two sided), then you'll detect it 90% of the time you run the test (on independent data).

"Some thought the game was harder with AI, others found it easier – but no one found it equally challenging."

So maybe it's just random? As in maybe some players randomly got an easier level in their first play than the second, and vice versa. I've played Don't Starve a little and each new game is very different.

Other points the article makes are more interesting, for example about players rating the AI version as more. It would be nice to see some confidence intervals, but I couldn't find the original research.