wch

When I run `npm audit`, it points me to a security advisory at GitHub. For example, for debug, it is https://github.com/advisories/GHSA-8mgj-vmr8-frr6 .

That page says that the affected versions are ">=0". Does that seem right? That page also says:

Any computer that has this package installed or running should be considered fully compromised. All secrets and keys stored on that computer should be rotated immediately from a different computer. The package should be removed, but as full control of the computer may have been given to an outside entity, there is no guarantee that removing the package will remove all malicious software resulting from installing it.

Is this information accurate?

I agree 100% with you. The author starts the essay with the very technique that he uses later on the essay: frame the conversation in his terms (that LLMs are _not_ inevitable), and then once that's done, anyone who disagrees can be dismissed.

The irony is that he's doing it, just in the opposite direction of those he disagrees with.

Some things _are_ economically inevitable, and it's wishful thinking to ignore them. For example, it is inevitable that solar will be the primary form of electricity generation in the coming decades (and I am thankful that this is the case). For those that were watching closely, this was clear several years ago, and it is even more clear today.

People whose fortunes are built on the fossil fuel industry might wish it were otherwise, and they might dismiss "solar inevitabilists" with the same technique as the author. But those people would be wrong.

If someone says that something is inevitable, they could be wrong. But they could also be right.

You should check out Positron, which is currently beta. It's created by Posit (the company formerly known as RStudio).

https://github.com/posit-dev/positron

A little context: this dive into Fortran is part of the excellent work George has been doing on WebR, to get R running in the browser. The R sources contain a fair bit of Fortran code, and I believe WebR originally used f2c to compile the Fortran to C first, before compiling that to wasm.

With the patches to LLVM Flang, WebR can be built with a real Fortran compiler.

I think George didn't want to say it directly in the blog post, but he has said that he's hoping that Flang would take his patches or implement better ones. That would be a win-win -- these patches wouldn't need to be maintained separately, and since unmodified Flang would be able to compile to wasm, it would benefit other projects out there that use Fortran.

https://docs.r-wasm.org/webr/latest/

This reads like it was written by AI.

Now that the poll is leaning toward "yes", Elon is adding conditions, namely, you have to invest your life savings. Lex Fridman volunteered to do it for free, and Elon replied:

One catch: you have to invest your life savings in Twitter and it has been in the fast lane to bankruptcy since May.

https://twitter.com/elonmusk/status/1604626103326253056

That looks fantastic. It also looks like you also have Pyright running in the browser. Is that right? I thought Pyright only supported running on NodeJS.

Did the original Wordle have a free license? I don't recall seeing license information anywhere.

I just want to say that I really enjoyed reading this article. It's among the clearest, most accessible writing about a technical subject that I've encountered in a while.

On human scales, the time dimension is much "bigger" than the space dimensions...

This is really interesting, and it made me wonder how to convert between space and time. I mean, one meter up is equivalent in magnitude to one meter forward, is equivalent to one meter to the right. Is _c_ the conversion between space and time? In other words, is 300 million meters equivalent in magnitude to one second of time?

Another serious question: if a stock has no voting rights and no dividends, why would it be worth anything at all? That is, what makes it different from a piece of paper with the company's name on it? If there are voting rights, then there's value because someone potentially could buy enough shares to meaningfully exert those rights, but if not, then what value do the shares have to anyone?

From my quick skim of the article, I think this is the key part:

These results imply that, after correcting for confounding variables (see notes under Table 3), those who were deficient at time of birth had on average 0.06 point higher gender-weighted SRS score, which corresponds to an increase of approximately one fourth of the s.d. (s.d. of the SRS in the total sample=0.23). When treating SRS as a dichotomous outcome, that is, comparing children with SRS scores above the suggested cutoff with children below the suggested cutoff, those who were vitamin D deficient at mid-gestation were 3.8 times more likely to be screened positive than those who were vitamin D sufficient at mid-gestation (see also Supplementary Table 4).

(SRS is the Social Responsiveness Scale)

There are people who waste money on toys, eating out, clothes and all that -- but those expenses pale in comparison to the really big costs:

- Child care (necessary if both parents work or there's just one parent)

- Larger apartment/house, in an area with good schools

My wife and I live in a large midwestern city. Prior to having kids, we had more than enough space in a 2-bedroom apartment. We had in the past (as students) lived comfortably on less than $20k each a year.

Putting an infant in a day care costs between $1200 and $1600 here. It drops a few hundred when the baby moves into the toddler room, and a few hundred more when the toddler moves into the older kids' room. With two kids in day care (an infant and an older kid), that still costs in the ballpark of $2000-2500 per month.

The 2-bedroom apartment is too small for 2 adults plus 2 children, plus visiting family members who help with kids, and moving to a bigger place in an area with good schools also increases monthly costs substantially.

Even if you're not spending money frivolously, having kids is expensive compared to not having them.

That's actually not true. He/she quoted the original two sentences. The parent post was later edited to add a third (middle) sentence.

It's a common misconception that for loops in R are slow. They're actually fast (around 20 million iterations per second on my computer). What can be slow is modifying data structures in particular ways using a loop. See http://rpubs.com/wch/46581

I found ages ago that setting up a mail server isn't super hard. What was difficult keeping it running reliably. Every time I would go on a trip, I'd worry that it would break down and I wouldn't be able to fix it. And I always worried that it was rejecting or spam-filtering mail that was important.

I agree that focusing after the fact has limited appeal, but there are some interesting long-term ramifications of this technology.

Today, much of lens correction are done in hardware, by adding more lens elements, which makes the hardware larger, heavier and more expensive. Some corrections can be done in software these days (barrel/pincushion distortion, chromatic aberration), but not all.

What they've been able to do with this camera is move a lot of those corrections into software, allowing them to use a simpler, cheaper lens with a very large aperture. The constant F/2 aperture is unheard of in a lens with such a large zoom range.

Besides aberrations, there's another issue with using traditional large-aperture lenses: focusing takes time, and it's not always accurate. With the Lytro, I imagine that you can take pictures instantly without needing to focus, and you can get perfect focus in post-processing.

From what I've seen in the sample gallery, image quality isn't great compared to regular camera/lens combinations in the same price range. This camera might not be for everyone, but I really hope the technology continues to improve and eventually become mainstream.

This post gives off a really bad brogrammer vibe.

And I hope these aren't developers they were hiring and firing. The average salary of these bad hires was about $57,000 per year. (The total was $70,000, divided by 6 people who stayed an average of 75 days.)

Farecast got bought by Microsoft and was used in Bing Travel, but it seems to be gone now. According to Wikipedia (http://en.wikipedia.org/wiki/Bing_Travel), "As of January 2014, the fare prediction feature has been removed."

You're getting "false positive" with the method you've chosen, but it's not a method that would be accepted in a scientific paper as evidence for an experimental effect. Maybe your method is more appropriate for, say, a machine learning context, but it's not what would be used in a paper like this.

First, the statistical tests used for these experiments don't make use of Bayesian stats, so the prior 50%-loaded probability simply isn't factored in. The standard is to use null-hypothesis testing, which says roughly, that if the null hypothesis is true -- that is, if there is no actual difference between the populations (experimental groups A and B, for example) -- what is the probability that you'd see a pattern like the one observed in the data. And the tests take sample size into account in calculating this probability.

If you throw the die once, the test that you'd use here (Chi-square) would _never_ give you a false positive, that is a p-value of <.05. With small samples, there is too little power to get a the requisite p-value. (And I'll note that Chi-square is one of the tests used in these papers.)

There's a whole other debate about whether p-values and null hypothesis tests are the right thing to use, whether the standard 0.05 threshold p-value is small enough, whether Bayesian stats should be used, etc. These are legitimate issues. But they're separate from the claim that small samples will increase the likelihood of a false positive.

I agree that there is a general problem that researchers go on fishing expeditions with their data. But having a small sample size doesn't make it any more likely to find a false positive.

We've already agreed that a small sample size doesn't make it any more likely to find a false positive for a given hypothesis. This is true for H1, H2, H3, etc., where each of these is a hypothesis. Therefore the aggregate effect of testing N different hypotheses is that you're no more likely to find a false positive with a small sample size vs a large sample size. You are more likely to have false negatives with small samples, though.

Those samples might be on the smaller side, but they're not "really small". And no, a small sample size doesn't mean that there's a high chance that there's a false positive. What it does mean is that there's a higher chance for a false negative - that is, the experiment fails to detect a real effect. That's what the standard statistical tests will do.

In addition to the _total_ debt vs. _median_ income issue that others have mentioned, this is a classic case of how 2 y-axes on a graph can mislead. Or it can be used to tell any story you want. Thiel's graph "shows" that debt is taking off and way exceeding income. Here's a different version of the same data which "shows" that debt and income are converging: http://i.imgur.com/OBCyfVb.png

Is it possible to use more than one device to control an outlet? As far as I know, Bluetooth devices can only be paired to one other device at a time. If that's the case here, then the outlet is tied to (for example) one phone -- but I might have a couple phones, tablets, and computers from which I'd want to control the outlet.

> A good reminder to not get stuck in local fad maximums.

That's a good goal - but how do you know in advance whether it's a fad (like Zynga) or the real deal (like Facebook)?

Experimenting on yourself with replacing food with shakes is one thing. I find that an interesting project, and I'm sure many others did as well.

On the other hand, selling this stuff with claims like, "Soylent is perfectly balanced and optimized for your body and lifestyle, meaning it automatically puts you at an optimal weight, makes you feel full, and improves your focus and cognition," is another thing completely. It sounds like a total scam, especially since they don't even say what's in it.

Nice! I've adapted your code and made a version that uses ggplot2. It also uses memoise() so that each data set only needs to be downloaded once per R session.

    # Memoize read.csv so each data set only needs to be downloaded once
    require(memoise)
    readcsv <- memoise(read.csv)

    fluPlot2 <- function(country="ca", regions="Nova.Scotia") {
      require(ggplot2)
      require(reshape2)

      url <- sprintf("http://www.google.org/flutrends/intl/en_us/%s/data.txt", country)
      d <- readcsv(url, skip=11, header=TRUE)
      d$Date <- as.Date(d$Date)

      # Convert to long format
      dl <- melt(d, id.vars = "Date", variable.name = "region")

      # Select regions of interest
      dlsub <- subset(dl, region %in% regions)

      ggplot(dlsub, aes(x=Date, y=value, colour=region)) + geom_line() +
        theme_bw()
    }

    fluPlot2("us", c("Minnesota", "California"))

It sounds cool, but I wonder how reliable this method is. What's the false positive and false negative rate? Errors here would have real consequences -- for example, according to the article, it was used as a crucial piece of evidence in putting several men behind bars for decades.

It crashed for me also. I wonder if it's because I have the offline gmail app for Chrome installed?

In your R version, you don't need the call to `which()`, so you could do this instead:

    dfrm$height[substr(dfrm$name,1,1) == "D"]

    subset(dfrm, grepl("^D", name), select = height)