Echoing others, Chess engines certainly aren't a solved problem! In fact there are a lot of niches that are absolutely starving for effort. Ones I'm interested in are related to Chess variants and puzzles.
Fairy-Stockfish is a fork used by LiChess for the variants on the site, but it can now play a multitude of games from Xiangqi (Chinese Chess) to Shogi (Japanese Chess) to a crazy modern variants. There's a variety of tools to train new neural nets for these variants, generate opening books, puzzles, etc. You can play some of them on PyChess (pychess.org). These are projects basically run by a couple people with huge backlogs of bugs and feature requests. An enthusiastic developer can easily get involved! Or just enjoy playing different variants and getting involved with the player community.
I've heard this referred to as either Placement Chess or Bronstein's Chess (after the GM who reportedly suggested it). It's available to play on https://pychess.org, where it's called Placement Chess.
Lichess also includes a piece set similar to this, although with the notable difference that the Knight is shaped like a greater-than symbol and the pawn is a small square.
The kanji on the pieces adds a significant potential barrier. I started with the piece set that had little directional pips on them and different colors to help (it's the twelfth set in the piece set list when you go to the settings gear on Lishogi).
Absolutely! And I'm one of them :) Check out the Shogi Harbor community[1] and their Discord[2]. There's a lot of discussion there and they're very friendly.
For those interested in playing and learning Shogi, take a look over at the great site https://lishogi.org -- a fork of LiChess for Shogi that includes real-time and correspondence play, AI opponents, analysis, puzzles and more. One of my favorite variants, Kyoto Shogi, is also available. The website https://pychess.org also has a ton of interesting and unique variants both traditional (Makruk) and modern (Chennis!).
This is a little surprising since MNRAS's niche has been that it had no page charges. So if one was short on grant money then you went with monthly notices. That's pretty nice if (a) you have zero grant funding available or (b) want to prioritize conference travel for students/postdocs.
MNRAS is one of the three "main" journals for astronomy along with Astronomy and Astrophysics (A&A) and the Astrophysical Journal (ApJ and AJ). A&A is mainly Europe and ApJ is more mixed but with more North America. Along with Nature and Science for high-impact things and regional journals for Japan, Australia, and the US (PASP).
This is an implementation of a neat pencil-and-paper game from the 1980's in Eastern Europe called Virus Wars or Klopodavka (bedbugs).
In this variant of Klopodavka, players can add a bug or squish an opponent's along cells adjacent to a path of friendly cells connected to the player's base. Players get six actions per turn and a player loses when no more valid moves are available. Although the AI is limited, it's an interesting game of territory and thinking about how one's own bugs will be used against oneself.
The code can be found at https://github.com/ptupitsyn/klopodavka-rs and was developed by Pavel Tupitsyn, who is also the maintainer of Apache Ignite according to his GitHub bio.
I don't think cursive is necessarily faster. But I do recall studies that showing what you're talking about: that a person's developed hybrid printing (printing with personal ligatures) are just as fast if not faster than full cursive.
What cursive does appear to be useful for historically is writing for long periods without tiring. Cursive written in the older Spencerian script fashion is written with big movements of the wrist and arm along the main diagonal path with little finger movement. The wrist and arms tire less quickly than fingers. Some professions in the 1800s like had to write all day, so it was quite useful and worth the significant training costs of learning such penmanship skills. They also had to write with pens that didn't write equally well in all directions.
And of course other fancy scripts likely just signalled class status similar to being able to recite Homer (in Greek) on command.
My understanding is that the first conclusive proof that the Earth rotates was the astronomical observation of the aberration of starlight[1] in ~1730. The rotation of the Earth causes about 20 arcseconds of deflection from a stars known position, which is pretty large. The phenomena itself is due to relativistic beaming, and lead to one of the first measurements of the speed of light along with Roemer's earlier work using the moons of Jupiter as a clock in ~1680.
Just changing my search habits helped a lot in in finding results quickly and managing the SEO spam for both DDG and Google.
I'm using Julia for a project right now, and answers are almost always found in either the Julia Discourse forum or Stack Overflow. By just searching either of those directly (plus more carefully reading the official documentation) I have found what I'm looking for much more easily.
The situation is more difficult with Python because the community aspect is more dispersed. But because my questions almost always have to do with using the standard library of external libraries, I've found devdocs.io to the be much simpler and more direct, especially because it runs offline. It covers >90% of cases for me with the docs for numpy, scipy, matplotlib, pandas, and base Python.
Maybe SEO spam forcing us back to RTFM will actually be a force for good?
Light pollution from satellite swarms is an economic externality. Any company can dump sewage in the river (and maybe the sewage is necessary for progress!) but the private sector profits by consuming a shared resource.
For the night sky and astronomy the externality is not so great in the grand scheme, but if unmitigated it does mean that there will be fewer discoveries per taxpayer dollar sent to the NSF/NASA. I'm a postdoc in astronomy and am all aboard Starlink-style networking. But it also seems fair to me that SpaceX should be the party that's responsible for treating that externality with light pollution, or else it's the same old "privatize the gains and socialize the cleanup." We've already effectively achieved this for radio astronomy by regulating protected bands in the spectrum for passive listening.
To anyone who does not get the irony in the above, this already exists as the Tenth Amendment. It is effectively ignored because it is considered legally redundant with the Constitution itself.
While I totally agree with the point you're trying to make, to be fair to the Byzantines, having a continuous state structure for a near millennium (500-1200 CE) is actually the near opposite of ossification leading to speedy collapse.
To Western Europeans coming from the very loose state structure of feudalism, the fact that the Byzantine Empire was effectively a modern state (similar to ancient China) was practically incomprehensible to them. Hence the phrase! On the flip-side, the level of institutional capacity let the Romans and Byzantines ride out incompetent Emperors fairly well compared to feudal states that would quickly disintegrate with one bad roll of the "off-spring lottery".
> Community and advertiser feedback made clear that we need to offer more ways to control the content that’s recommended as well as where ads appear. So, we’re introducing a new category: Pools, Hot Tubs, and Beaches. If you have chosen swimwear that is allowed under the “Swim and Beaches” contextual exception to our standard Nudity and Attire policy, you should stream into the Pools, Hot Tubs and Beaches category.
People set up kiddy-pools in their room with a bit of water to get this "contextual exception" for the express purpose of streaming suggestive content. Why not just remove the exception and let people stream sexually suggestive content and make a separate category for that without a ridiculous "beaches and pools" euphemism?
Start with a small set of binoculars! They are critically underrated. I think the best part of amateur astronomy is knowing the sky in an intuitive way. With a pair of 7x35 binos (~40 USD online) and a dark site you can see a ton of stuff and they're dead simple to use. And you'll still get a lot of use out of them as a finding-aid with larger aperture equipment. I keep a small set in my car just in case I'm ever at a dark site.
Even if you live in a heavily light polluted city, you can do something astronomy-adjacent with binos: satellite hunting! HeavensAbove[1] has a really nice free android app for it. Satellites are mainly visible at dusk or dawn, and dusk is a lot easier than staying up late depending on your schedule. Plus it's like a race, they may only up for less than a couple minutes, so you have to have a keen sky-sense to pick them out. It's cool being able to know what they are too -- "huh, that's some Soviet rocket booster from the 70's"!
That makes a lot more sense that many transactions are included within a block. So effective transaction fees are likely in the few cents USD range.
I've read that one issue with the latency of Bitcoin transactions is that many of the large miners from China have network issues. Would raising the block size to 10-100MB cause significant latency issues for transactions even if one could fit a lot more transactions into a single block?
Can someone explain how transactions are not insanely cost prohibitive for Bitcoin with power consumption figures like those? Those numbers would suggest something like >20 USD. Is it because most transactions in practice actually happen internally on exchanges that avoid putting every individual transaction on the blockchain?
I have limited knowledge of Bitcoin so maybe not quite "explain like I'm five"... but close. :)
Where people live just meant that choosing to distort the equator. This does not seem like a useful map to use if you are interested in Africa or South Asia! It's a tradeoff for all projections.
However, I don't want to go too hard on these guys. Reading the arXiv paper, I think it's clear that they wanted to make a cool set of disks for the Solar System. They actually look really cool for Jupiter and Saturn with their equatorial cloud bands and polar vortices! And centering on the poles gives a nice symmetry with the Earth and Mars with the ice caps.
They also show the version of Earth with the projection centered on the Western and Eastern hemispheres and the distortions are a lot more obvious (e.g., Alaska, Canada, and Australia are pretty bad). I can see why they like the North/South pole version.
Fairy-Stockfish is a fork used by LiChess for the variants on the site, but it can now play a multitude of games from Xiangqi (Chinese Chess) to Shogi (Japanese Chess) to a crazy modern variants. There's a variety of tools to train new neural nets for these variants, generate opening books, puzzles, etc. You can play some of them on PyChess (pychess.org). These are projects basically run by a couple people with huge backlogs of bugs and feature requests. An enthusiastic developer can easily get involved! Or just enjoy playing different variants and getting involved with the player community.