3 ms·

So your bufferbloat curve comment refers to how congestion doesn't depend linearly on symbol rate, e.g https://superuser.com/questions/1324956/is-bufferbloat-t

So your bufferbloat curve comment refers to how congestion doesn't depend linearly on symbol rate, e.g https://superuser.com/questions/1324956/is-bufferbloat-triggered-on-a-certain-speed https://superuser.com/questions/1324956/is-bufferbloat-trigg...? Or is there more to it (like hidden Shannon assumptions not priced in with the chiprate e.g https://news.ycombinator.com/item?id=30130091 https://news.ycombinator.com/item?id=30130091 https://phys.org/news/2024-06-cooperation-evolve-limited-payoff-memory.html https://phys.org/news/2024-06-cooperation-evolve-limited-pay... ) Chemical clocks can get a bit crazier-- there can be much randomness, but the number of equivalent ahem pathways can be few. https://pubs.rsc.org/en/content/articlelanding/2015/cp/c5cp02572a https://pubs.rsc.org/en/content/articlelanding/2015/cp/c5cp0...

082349872349872 2y ago
"Tit for Tat" only requires storing one bit, for the last interaction. (its robust-to-noise variant, "Tit for Two Tats" still only requires constant memory) (I think I've talked about bufferbloat on HN before, but my Algolia-fu is lacking today) In this particular parent, I'm not referring to bufferbloat, but rather the mechanism "chipping", by which spread-spectrum communication can hide near, or even below, what more traditional signal processing might consider the "noise floor". Jiang Ziya gets a good start on the idea in §25: "Secret Letters" of the "Dragon Way" (httphttps://zh.wikisource.org/wiki/六韜#陰書第二十五 https://zh.wikisource.org/wiki/六韜#陰書第二十五) in which he suggests sending sensitive messages by sending 1/3 of their characters each over three different channels. The birds (鸟语) use frequency sweeps (chirps) to provide noise protection: most noises either occur across nearly all frequencies but are localised in time (a "pop") or occur across long stretches of time but are localised in frequency (a "hum"); by changing frequency with time one minimises interferences with either of those noise sources. Finally, frequency-hopping proper involves sending multiple chips for each bit, on a nearly uncorrelated (prng-drived) schedule of frequencies. (because we have sign as well as frequency, even on a single frequency swapping between sign conventions allows one to extract signal from substantial amounts of noise. A Dicke switch would be the scientific method in microcosm, hopping relatively quickly between signal and control) EDIT: Mao Basic Tactics (1937) mentions multiple redundant messaging, but only as a repetition code, not even as fancy as Jiang's scheme. [Mao evidently had the problem of composing fighting forces capable of independent action with peasants who could read fewer than 20 characters! Yikes! (I mean, not that I'm much better, but I'd guess by analogy with kanji I probably have 20 hanzi already just in passing?)] vinnyvichy, any chance your baidu-fu would be enough better than mine to find either that or On Guerrilla Warfare (1937) in the original hanzi?
- vinnyvichy 2y ago
  https://www.marxists.org/chinese/maozedong/marxist.org-chinese-mao-193805.htm https://www.marxists.org/chinese/maozedong/marxist.org-chine... (1938-05) "由此，游击战争的指挥原则，一方面反对绝对的集中主义，同时又反对绝对的分散主义，应该是战略的集中指挥和战役战斗的分散指挥。" Glad to be able to be meatgpt for now
  - 082349872349872 2y ago
    Thanks; that's 1938* 《抗日游击战争的战略问题》but I'd been hoping for the 1937 work,《游击战》, which even in english only seems to appear in various US translations as well as a "collected works" that was printed in India, but is not canonical according to China? When I try baidu-ing I get plenty of hits for Guevara's title but none for Mao :( * I think en.wikipedia.org is just plain wrong here but maybe they consider them two editions of the "same" work?
    - vinnyvichy 2y ago
      https://zhuanlan.zhihu.com/p/512552544?utm_id=0 https://zhuanlan.zhihu.com/p/512552544?utm_id=0 says 1938 might be what they meant. Might have to get to a physical archive to make sure.
      - 082349872349872 2y ago
        > 现在其中文版已经找不到了 Yeah, Griffith's translation is what I keep turning up, no matter what character set; maybe if I ask Fort Meade really nicely "The Dragon Lady" would be willing to samizdat me their version of the original?
        vinnyvichy 2y ago
        But but but that the Griffiths translated "contents" line up nicely with the 1938 hanzi contents doesn't excite you? https://banned-historical-archives.github.io/articles/43e71003d4 https://banned-historical-archives.github.io/articles/43e710...

082349872349872 2y ago

"Tit for Tat" only requires storing one bit, for the last interaction. (its robust-to-noise variant, "Tit for Two Tats" still only requires constant memory) (I think I've talked about bufferbloat on HN before, but my Algolia-fu is lacking today) In this particular parent, I'm not referring to bufferbloat, but rather the mechanism "chipping", by which spread-spectrum communication can hide near, or even below, what more traditional signal processing might consider the "noise floor". Jiang Ziya gets a good start on the idea in §25: "Secret Letters" of the "Dragon Way" (httphttps://zh.wikisource.org/wiki/六韜#陰書第二十五 https://zh.wikisource.org/wiki/六韜#陰書第二十五) in which he suggests sending sensitive messages by sending 1/3 of their characters each over three different channels. The birds (鸟语) use frequency sweeps (chirps) to provide noise protection: most noises either occur across nearly all frequencies but are localised in time (a "pop") or occur across long stretches of time but are localised in frequency (a "hum"); by changing frequency with time one minimises interferences with either of those noise sources. Finally, frequency-hopping proper involves sending multiple chips for each bit, on a nearly uncorrelated (prng-drived) schedule of frequencies. (because we have sign as well as frequency, even on a single frequency swapping between sign conventions allows one to extract signal from substantial amounts of noise. A Dicke switch would be the scientific method in microcosm, hopping relatively quickly between signal and control) EDIT: Mao Basic Tactics (1937) mentions multiple redundant messaging, but only as a repetition code, not even as fancy as Jiang's scheme. [Mao evidently had the problem of composing fighting forces capable of independent action with peasants who could read fewer than 20 characters! Yikes! (I mean, not that I'm much better, but I'd guess by analogy with kanji I probably have 20 hanzi already just in passing?)] vinnyvichy, any chance your baidu-fu would be enough better than mine to find either that or On Guerrilla Warfare (1937) in the original hanzi?

vinnyvichy 2y ago
https://www.marxists.org/chinese/maozedong/marxist.org-chinese-mao-193805.htm https://www.marxists.org/chinese/maozedong/marxist.org-chine... (1938-05) "由此，游击战争的指挥原则，一方面反对绝对的集中主义，同时又反对绝对的分散主义，应该是战略的集中指挥和战役战斗的分散指挥。" Glad to be able to be meatgpt for now
- 082349872349872 2y ago
  Thanks; that's 1938* 《抗日游击战争的战略问题》but I'd been hoping for the 1937 work,《游击战》, which even in english only seems to appear in various US translations as well as a "collected works" that was printed in India, but is not canonical according to China? When I try baidu-ing I get plenty of hits for Guevara's title but none for Mao :( * I think en.wikipedia.org is just plain wrong here but maybe they consider them two editions of the "same" work?
  - vinnyvichy 2y ago
    https://zhuanlan.zhihu.com/p/512552544?utm_id=0 https://zhuanlan.zhihu.com/p/512552544?utm_id=0 says 1938 might be what they meant. Might have to get to a physical archive to make sure.
    - 082349872349872 2y ago
      > 现在其中文版已经找不到了 Yeah, Griffith's translation is what I keep turning up, no matter what character set; maybe if I ask Fort Meade really nicely "The Dragon Lady" would be willing to samizdat me their version of the original?
      - vinnyvichy 2y ago
        But but but that the Griffiths translated "contents" line up nicely with the 1938 hanzi contents doesn't excite you? https://banned-historical-archives.github.io/articles/43e71003d4 https://banned-historical-archives.github.io/articles/43e710...