Daniel Lemire's blog

, 9 min read

Science and Technology links (December 4th 2021)

9 thoughts on “Science and Technology links (December 4th 2021)”

  1. Maynard Handley says:

    Hi Daniel,,
    Have you read https://astralcodexten.substack.com/p/book-review-lifespan ?
    Seems very much your kind of thing…

    1. I deliberately ignore Sinclair.

      1. Maynard Handley says:

        The review is a nice summary of the pros and cons of his stance. I guess you know it already, but it was new to me.

        1. Reading his book can’t hurt. I will not review it here.

      2. Roc Reguant says:

        Any particular reason? Too much marketing? Hidden agenda?

        1. Too much marketing.

  2. José Duarte says:

    I used to be so excited by these different companies designing their own chips. Amazon with G1 and G2, Microsoft Azure with their FPGA accelerators, the Google/YouTube custom video transcoding silicon, etc.

    Now it’s starting to feel like people inventing a new type of stapler while on a sinking ship. Well, I guess more like inventing amazing chips while on a sinking ship, or while working for the USSR or something.

    The explosion in strange censorship is deeply troubling. When AWS kicked Parler off their cloud, I was stunned. It’s still not clear how Parler is different from Twitter or Facebook, in terms of percentage or number of violent users, or supporters of violence or what have you. Apparently you can’t have a social network in America unless you’re willing and able to vigorously monitor and censor your millions of users, according to some unknown, unstated standard. I had no idea that this was the situation, that you literally aren’t allowed to compete with the incumbents unless you can spend millions of dollars a year just on censorship ops.

    So now I don’t care about AWS and their chips. I think it’s sad that any engineers are willing to work for Amazon under these conditions, where mundane non-leftist discourse is apparently not allowed. At least Amazon isn’t yet censoring non-leftist books, though there are constant efforts to get them to do so… And AWS’s ridiculous propaganda endorsing the disturbing BLM movement/violence was so offensive that I diverted major purchases to other firms.

    Since I wouldn’t be able to use the new chips if I wanted to host something as mundane as social network that doesn’t monitor and censor its users, their performance is moot. Those guys should work elsewhere.

    1. Now it’s starting to feel like people inventing a new type of stapler while on a sinking ship. Well, I guess more like inventing amazing chips while on a sinking ship, or while working for the USSR or something.

      The USSR was pretty bad at inventing and producing new computer chips.

      I do not think you need to care particularly about Amazon, Apple or any of these guys to find the technology that they are producing interesting. I am not advocating people use Amazon products or services (I am neutral).

      Regarding business decisions, I think that it is good to favour vendor independence: be very concerned that your business depends on a single provider. I have seen entire startups wiped out because Facebook changed the terms of service.

      1. José Duarte says:

        I agree about Amazon, since AWS offers valid and useful services. I guess I still like to read about their technology and engineering. Netflix too. Facebook is actually the depressing one for me – it’s such a waste of engineering talent to build a massive compute infrastructure so that idiots can post selfies. Same with all the AI people there building better ad targeting. Why? Why not do something else?

        On the censorship front, Facebook recently crossed over into tagging new scientific research as “false” for the first time we know of. They partner with leftist activist groups to censor “misinformation”. One of those groups is called Climate Feedback, and they did something we’ve never seen: they declared a new paper “false” without even reading it. I’m not sure if they’ve ever done it before, or this was the first time. The paper argued for a new estimate of solar forcing in Northern Hemisphere surface temperatures, an upward revision (https://arxiv.org/abs/2105.12126). It’s a long story, but the activist at Climate Feedback thinks that humanity already knows everything we need to know about climate change on this planet, and even specifics like solar forcing, even focused on specific hemispheres. He thinks that a very young and small science is “settled” maybe 20 years into its current methodological framework (knowing the recent history of computing would also argue against such silliness – climate scientists have only had access to powerful modeling computers for 20 years at most, less for some teams). It’s a stunning situation, true unfalsifiability and preemptive rejection of new scientific research if the findings seem to contradict one’s existing, politically driven beliefs. All sorts of things didn’t occur to him, like why climate scientists would be conducting any research on solar forcing if it’s settled, why a peer reviewed journal would publish revised estimates of solar forcing or any other aspect of climate if such estimates were somehow knowably false, why climate science journals still publish papers and continue to exist, and so forth. This was a watershed moment. Facebook is now censoring new science, tagging it as “false”, and reducing the reach of users who post it.