It's a bit like an abacus: the motion of beads on each individual rod may not seem to make much sense, but there is a meaning to the calculation taken as a whole.
(Hmm: presentation does a great deal for providing meaning; imagine a "write-only" tetris game where instead of presenting the current state of the playfield, completed rows stayed on the display and dropped pieces stretched through them to incomplete rows. I think that'd be much more difficult than the traditional view)
"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) 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?
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?
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?
I don't doubt that it's an accurate translation (Griffith retired from the USMC and did a doctorate in chinese military history before[0] attempting it); I'm just curious[1] how various phrases from the translation were phrased in the original.
[1] the only other chinese text I've looked closely at has been bits of the Mawangdui Tao Te Ching, and I'd guess using that as a reference style would be like attempting to write contemporary english based on the King James bible.
I might be misreading the signal in your messaging as of your frustration that 1938 text
I posted ("//..banned-...") isn't the original source of Griffith translation
Here is the first para:
"In a war of revolutionary character, guerrilla operations are a necessary part. This is
particularly true in war waged for the emancipation of a people who inhabit a vast nation.
China is such a nation, a nation whose techniques are undeveloped and whose communications
are poor. She finds herself confronted with a strong and victorious Japanese imperialism. Under
these circumstances, the development of the type of guerrilla warfare characterized by the
quality of mass is both necessary and natural. This warfare must be developed to an unprecedented degree and it must co‑ordinate with the operations of our regular armies. If we
fail to do this, we will find it difficult to defeat the enemy. "
Okay you might rue that it isn't a scan, but I doubt the hanzi isn't a word for word transcription of the original. Unless.. you mean you want the traditional hanzi?
The one translation I find questionable is of the phrase 民众性 to "quality of mass", maybe appealing to Clausewitz?
EDIT: in fact if you click on 对比 you get scans of the original, but it crashed my mobile browser lol..
I'm ruing that we have 1938 in both languages, but 1937 only in english translation (although I have every reason to believe that translation has, when choosing between fidelity and beauty, opted for the former), and I don't believe 1937 and 1938 are the same composition (or at least if 1938 is an expanded 1937 it also clearly has deletions).
[traditional vs simplified is not an issue here, although I note that 丑 (note there are sibboleths even in the simplified strokes!) as simplified 醜 loses a potential flight of fancy: does the existence of unitary ghosts imply that subghosts and ghost-quotients are two different structures (in general: for abelian hauntings they coincide?), for the latter may be non-unital? This would have important repercussions for the kernels of mappings between ghosts...]
Hard to do this on a mobile phone so maybe point me to suspected insertions/deletions in the 1938 translation as you seem to already have a list of them.
(Hmm: presentation does a great deal for providing meaning; imagine a "write-only" tetris game where instead of presenting the current state of the playfield, completed rows stayed on the display and dropped pieces stretched through them to incomplete rows. I think that'd be much more difficult than the traditional view)
Lagniappe: https://news.ycombinator.com/item?id=40714981
(Sorry for interrupting your regular HN? Maybe we ought to bump up the https://en.wikipedia.org/wiki/Chip_(CDMA)#:~:text=The%20chip.... rate and attempt to dive beneath the noise floor again?)