trendia

Or he had some really expensive meds and bad insurance. That's pretty unlikely though.

Link is dead for non FT subscribers.

(Or at least for those who the FT thinks would pay for a subscription)

I don't think you can even get a reliable 10 gbit PCI card for under $200 USD.

"Imagine you design a website, and then Google changes their ranking algorithm to provide preferential treatment to AMP pages. You see that your ranking has decreased, and so you implement AMP.

"A few months later, Google announces that they will provide preferential treatment to websites that have progressive web apps, and you see that your website falls in the rankings to the second page. Once again, when Google asks you to jump and dance like a moron, you flail your arms and repeat that you are not worthy."

This is pretty similar to what happened to me. I was implementing an algorithm that had multiple parts and the data would arrive asynchronously.

The first time, I wrote the multipart algorithm using a functional style to encode each state. As the algorithm grew, it became more and more difficult to reason about and managing dependencies between the parts was harder, not easier.

When this became unwieldy, I rewrote it. However, when I wrote it the second time, I wrote the algorithm in a way that made it appear imperative. Everything was linear, starting with part B, then part C, etc. A small backend managed state, but you could read through the algorithm code and it almost seemed like it would run synchronously.

The second method was easier to read, debug, and modify. I will probably not go back to the functional method anytime soon.

A video would be worth 1,000 articles.

Totally irrelevant to this article, but you are technically correct.

Let X = "angle"

Let Y = sin(X)

Then X has a causal relationship with Y, but Corr(X, Y) = 0.

I doubt that's what you're saying, though!

On the one hand, her account is very inspirational to people who might think that learning physics is out of reach or they're too old to do it. But on the other hand, her case is very out of the ordinary -- I think a lot of us took longer than 1.5 years to get through undergraduate physics.

According to Susan's account, it took her 1.5 years to go from no college-level math to a full college-level physics curriculum. At the school I studied at, the Calc series (I, II, and III) takes 1.5 years. The intro physics (I and II) takes a full year, and the upper physics take at least a year (Mechanics I & II, E&M I&II, Relativity I&II, Quantum I&II, Lab I&II). (And you had to take intro physics before higher-level physics, and most courses had dependencies.)

That's approximately 45 credit hours. If you took 18 credits a semester, that would take at least 2.5 years. And those are all tough courses -- no 3 credit gen ed "filler" courses. Even at 15 credits a semester, that's still 3 years.

So, to get through all of those courses in 1.5 years would be like learning about 30 credit hours / semester. And add in the physics electives, like thermo, waves, statistical mechanics, quantum field theory, additional math courses, etc... that's a lot on top.

I guess my point is that if Susan's account is discouraging, note that a lot of people take far more than 1.5 years to get through undergraduate physics.

If removing "features" improves performance, one has to wonder how much overhead those features have.

If I remove the UI/window manager from Linux and run software from the command line, I don't see a massive performance boost. That's not to say I would run a server with the UI, but it's not like I'm seeing a 100% improvement or anything. Thus, there must be some difference between the acceptable overhead of Linux vs. Windows.

It just makes me think that Windows continues to add features as long as that n+1 feature doesn't slow down the PC, and thus doesn't stop until the PC is slowed down. Thus, every Windows OS will have a maximal amount of overhead.

Oh sorry, I wasn't clear about that. I run it first where every possible module is built 'M', and then once the system boots up, run lsmod to see which modules are loaded.

Then, I take that list of modules and set them to "Y", and disable everything else. That way, needed libraries are statically linked and unneeded libraries aren't built at all.

I'd also look into Buildroot [0]. I'm currently using Buildroot for a project and the whole distribution (including libraries and executables) is under 40MB. Since I am using a fixed dev board (i.e. peripherals are not going to change), I used lsmod to detect which drivers are needed and only build those, which really shrinks the kernel.

Buildroot also includes a cross compiler, so that you can rebuild the entire toolchain, kernel, and libraries in one go.

[0] https://github.com/buildroot/buildroot

Swimming is great because it doesn't reward effort as much as it rewards form. I came to swimming from a running / cycling background, and I quickly realized that extreme exertion was not the key to getting better speeds. It's only when I started to consciously focus on form and accepted being slower during training, that I saw much improvement.

Also, congrats on the 1:07! I'm not even close to that speed yet.

Intel says, "Intel continues to work closely with industry partners to protect customers against the security exploits disclosed by Google Project Zero."

Linus Torvalds says, "They do literally insane things. They do things that do not make sense ... The patches do things that are not sane. WHAT THE F*CK IS GOING ON?"

There is an easy, quantifiable metric that uses this idea: the Brier score.

Essentially, if you make 100 predictions, the Brier score will measure how well you are able to make precise predictions. If you always estimate the odds that X will win to be 50/50, you won't do nearly as well as someone who correctly predicts the outcome each time.

https://en.m.wikipedia.org/wiki/Brier_score

The authors only looked at weight and BMI, and not bodyfat % or lean mass, which are much better indicators of health than weight or BMI alone.

Someone who is 6' and 200 lbs could either be a body builder or a coach potato, and they'd have the same BMI.

I had the opposite experience. I asked Google:

"Hey Google:, whatis the score of [University] basketball game?"

Even though the game was going on, Google told me about the women's basketball game the following day. It took about 5 min before I could get Google to give me the results I wanted.

I have had a lot of results like this--Google becomes useless. It's faster for me to search for it by hand.

Judea Pearl wrote a whole book about proving causality (and when it can't be proven).

A rough summary is available as a YouTube video:

https://m.youtube.com/watch?v=HUti6vGctQM

> we should easily see it.

Note, I'm not saying there is an effect, but it's not quite so easy to say we should have found it if it exists. First, there is a delay between radiation exposure and cancer risk (up to ten years).

Second, if all population centers experience the same increase in radiation exposure, then it is very difficult to use radiation as an independent variable to measure its impact on cancer.

For instance, it took decades before anyone was able to show that nuclear testing caused an increase in deaths, and that was only possible because the distribution of radioactive isotopes was not uniformly spread across the US -- some states (especially in the midwest) received far more radiation than others (e.g. Western California or Louisiana).

In contrast, if electromagnetic radiation from cell towers is relatively uniformly spread for the same population density (e.g. New York is comparable to Los Angeles, and rural Washington is comparable to rural Missouri), then it will be rather difficult to separate cancer risk from electromagnetic radiation from other risks.

I listen to podcasts or audiobooks while commuting, and I figure it's roughly equivalent to sitting on my couch reading. I commute 40-60 minutes daily, so it's pretty easy to get through long works that way.

> "Python has been used by hundreds of happy users"

Hundreds!

> "Please don't ask me to mail it to you -- at 1.76 Megabytes it is unwieldy at least..."

I can not even figure out if this means email or snail mail.

It's too bad the author didn't mention Wyden until the 5th paragraph. The author puts "Senator" in the title, when "Senator Wyden" would be just one more word.

The US won't be able to force these rules on foreign companies selling outside the US. e.g. Huawei, Samsung, and Sony are not going to backdoor their phones for customers in Asia / Europe. So, corporate / governmental customers in those countries are going to shy away from buying American phones... I can't see how this is a good thing for American companies.

> Not a lawyer. Would welcome a legal opinion.

I would highly recommend reading at least the Wikipedia article on "Disparate Impact."

https://en.wikipedia.org/wiki/Disparate_impact

> But if you're searching enough to eat up Google's bandwidth, they're paying for that data and they're under no obligation to keep serving you as a client

Do you not see the irony?

I can somewhat understand where you're coming from if the person doing it only wants something that works.

But for those who would be willing to risk breaking a few things to try something out, building a kernel is a worthwhile effort, and the meltdown / spectre bugs provide a perfect excuse to do it.

(Copying my instructions from another post).

If kernel 4.4 doesn't work, I recommend compiling the 4.15 kernel. (Note, however, that you may need to apply a patch to NVIDIA drivers).

I've done this on Ubuntu 16.04 LTS, 17.10, and Debian 8 so far this week. To compile, set CONFIG_PAGE_TABLE_ISOLATION=y. That is:

    sudo apt-get build-dep linux
    sudo apt-get install gcc-6-plugin-dev libelf-dev libncurses5-dev
    cd /usr/src
    wget https://git.kernel.org/torvalds/t/linux-4.15-rc7.tar.gz
    tar -xvf linux-4.15-rc7.tar.gz
    cd linux-4.15-rc7
    cp /boot/config-`uname -r` .config
    make CONFIG_PAGE_TABLE_ISOLATION=y deb-pkg

I think you mean "debt" -- I first read your comment as "code department" and was trying to figure out what that was.

It took about an hour for it to find the offset for me.

I think that the page isolation slows it down, even if it doesn't completely eliminate it.

The second test had something like a 0.05% success rate on my PC, and took over an hour to get a few dozen values read.

After trying this with the new kernel, I started up an AWS instance and ran the tests there. The first test (KASLR) succeeded within a few seconds, and the second test had a 100% success rate (read 1575 values in a few seconds).

"I knew I had made the right choice to stay out of cryptocurrencies when my taxi driver launched an ICO." -- Warren Buffet, sometime in the future

Linux 4.15 and the appropriate modules protect against the attack.

To test, set CONFIG_PAGE_TABLE_ISOLATION=y. That is:

    sudo apt-get build-dep linux
    sudo apt-get install gcc-6-plugin-dev libelf-dev libncurses5-dev
    cd /usr/src
    wget https://git.kernel.org/torvalds/t/linux-4.15-rc7.tar.gz
    tar -xvf linux-4.15-rc7.tar.gz
    cd linux-4.15-rc7
    cp /boot/config-`uname -r` .config
    make CONFIG_PAGE_TABLE_ISOLATION=y deb-pkg