There is a series of four fantastic videos about a successor to Thompson's tidal analyzer - the Michaelson's Harmonic Analyzer; here is the first video:
https://www.youtube.com/watch?v=NAsM30MAHLg
To me, the greatest appeal of analog computing is in representing equations in mechanical, visual, directly modifiable, 'hands on' forms. Having different concrete representations or implementations of abstract concepts makes them so much easier to understand. I've read somewhere that the technicians working at MIT with Vannevar Bush's mechanical differential analyzer quickly learned advanced calculus and "debugged" the math of university professors that would use the machines for various computations.
It is not the continuous nature of the variables that is so interesting, it is the concept of computing by analogy and the benefits to understanding that come with that.
Journals don't "do" anything, but the professors, PIs, postdocs, etc are evaluated for jobs or promotions based on the "quality" of their work approximated by the "quality" of the journals they were able to publish in. A Nature or a Science paper gets you a lot more than an arxiv preprint (in most fields, anyway). My best bet is we need a new way of evaluating scientific work, without relying on proxies like journal impact factors. Not sure what those might be, probably depends on the field.
I would say the title should be "reciprocal" of a vector. Right? The inverse of a function (f^-1) has an unfortunate notation equality with the reciprocal (x^-1), or multiplicatory inverse. Or am I wrong?
The whole point of statistics is dealing with those variable phenomena in psychology. Variability is the starting assumption. IMO, we need to find new methods of discovering non-variable and reproducibe phenomena. I like the theory that goals are relatively constant while means (behavior) are variable. The problem is to discover the goals.
Carbonara, though. Pasta, eggs, bacon, cheese and pepper. Pigs, cows and wheat came with the first farmers, some 6k years ago. Chicken around 2k years ago, but they could have used other eggs. Pepper much later, silk road trade. I imagine they could have had something like carbonara a long time ago.
>Yet he also made contributions larger than anyone will ever do by filling up note-taking apps and tracking how much time they spent on their breakfast with a stopwatch.
Maybe not the best example. Apparently, Newton took loads of notes, and also created handwritten indexes and contents lists, alphabetical, by topic.
If it is safety regulation you're talking about, that is not exactly a plus for the local population. Sure, it is a non-renewable source of power, but it doesn't produce CO2 after it starts to operate. Ideally, it could be even used to capture carbon from air.
I was just looking for smaller scale iron spreading projects, and apparently this was already done, with mostly positive effects, wikipedia seems to have good info: https://en.wikipedia.org/wiki/Iron_fertilization
There is also fertilization from natural sources of iron and other minerals that we can observe - the sands of Sahara that fly to the Atlantic and the Amazon basin, some vulcanic eruptions had high iron content, etc.
I suppose slowly increasing the scale of iron fertilization experiments could be a good direction.
> Authors found X. Seems interesting. Here's how they're interpreting it. Here's my speculation.
Agreed, that is mostly how it goes.
On a similar note, null-hypothesis testing is not a true and tried scientific method. It started sometime in the early 20th century. Aside from p-hacking and mistakes of that kind, there are many philosophical problems with the use of statistical inference itself.
> The equipment you get there is usually the result of a failed attempts by trained personal to fix things
Are there statistics on how many things were successfully fixed vs how many ended up in the maintenance center? If not, that is a possible survivorship bias, kind of.
I'm learning to solve differential equations with analog computer techniques. I don't have access to an electronic or mechanical analog machine, only simulations, but I find simulations quite flexible and enjoyable. There are a lot of great books from 50's and 60's on the topic.
> for scientific theories that actually have practical impacts, you don't have to just accept what scientists say on trust; you can look at their track record of correct predictions
What do you think of economic theories or theories behind psychotherapy? Lot's of real world impact, low confidence in experiments, imo
I've read the opposite from gwern - that nicotine is safe, and the carcinogens are mostly in the burning of tobacco products. He tested nicotine as a nootropic, apparently has significant effect.
I quit smoking with Carr too, some half dozen times. It would last for a few months, then I would start again after a few beers. The latest trick, good for two years now, is using nicotine-free rolling herbs (edit: things they give as filler to roll your joints, a mix of things like papaya leaves, tea, etc)
It is written in go and compiled. Go is responsible for high parallelism. The fact it is compiled is an advantage over javascript-based bundlers because they are re-compiled every time you start building the source code (did I get that right?).
Amazing collection! I'm learning about backprop in neural networks and I'm reading how that kind of optimization may not be biologically 'implementable'. Are non-gradient based optimizers an alternative proposal for biological learning? How do they compare to backprop?
If I understand correctly, at the moment if someone would want to do the 51% attack, they would need to spend energy for proof of work that would overpower the rest of btc the network, which is equal to expenditure of Argentina in kWh. A lot of people complain about this expenditure.