asciilifeform: ( but as it is, we also have a wild animal, in the mix, called miner, and its immediate ecosystem )

asciilifeform: given as i do not know a reason why it oughta be as low as 508 when going from ip stack mtu would be 1472.

asciilifeform: btw , esp for diana_coman , e.g. 1000 byte payload will be a 1028 byte total mass. (header)

asciilifeform: mircea_popescu: if yer using a linux that 'came with things', it's likely to have both

asciilifeform: a ty mircea_popescu

asciilifeform: i did have a tiny shred of 'hm maybe modern gear reassembles', but it was just that, a lark

asciilifeform: mircea_popescu: i can't fathom why kernelism would want a 4.9 specifically

asciilifeform: http://btcbase.org/log/2018-09-24#1853724 << incidentally ave1's cemented gnat is a 4.9 ☝︎

asciilifeform: it dun even switch on w/out a large fw turd, which i'm guessing mircea_popescu did not have

asciilifeform: mircea_popescu: i dun think i ever got even as far as this, with a radio nic made post-'08; yer ahead of me

asciilifeform: it's a pre-obummer box.

asciilifeform: 0 soldering, thing had a lid

asciilifeform: ( for folx who aint sov-entomologists : pavlik morozov was a kid who turned in own father to dekulakizers. for this, grandfather beheaded him with axe. and kid was made into an Official 'hero' )

asciilifeform: even per this 'logic', seems like there's a loose end -- so, ok, they sell, SELL, only. bags of benjies -- go in. ~then~ wat. ☟︎

asciilifeform: mighty lotta cheek, for folx who torment a 300 $ crate in tariff hell, to prattle about 'international marketing' ☟︎

asciilifeform: BingoBoingo: i've yet to meet the orc who did not say 'mine is a poor country.' after all, so far from miami.

asciilifeform: '"I am not happy with the dominance of Chinese contractors. In the first place, the money that they get from these contracts is externalised and all that they return here are meagre wages," said Edgar Syakachoma, himself a contractor. "Let the government also give us the contracts so that they benefit Zambians."' << hey, recall mircea_popescu's piece with the argentine anti-bus protesters

asciilifeform: then again we may already be in the midst of just such a thing, there's, what, ~20k distinct nodes operating, and overwhelming majority of these, are simply rubbish

asciilifeform: and theoretically if the cartel births '9000' new sybils, it will force the use of something like my algo, whether anyone wants to or not, the layer of sibylade would become undrillable in its absence . but this is admittedly a stretch. ☟︎

asciilifeform: but end result may end up being a net where even moar 'winner take all' than presently.

asciilifeform: Mocky: the current algo optimizes for nuffin at all. but the puzzler is, if we were to optimize for ~something~, could it create a problem. esp if the algo worx as i predict, and the nodes using it, end up forming an optimally short link to the miners.

asciilifeform: 'who' is a misconceived, somewhat, q, cockroaches dun have faces

asciilifeform: Mocky: the boojum is that mining in classical btc worx in such a way as to amplify small advantages into gigantic

asciilifeform: this is a potentially dangerous arse-mouth system. ☟︎

asciilifeform: and not merely because A,B,C reliably supply correct blox, but they also will have most reliable tx-eating. so they will get sent all new tx in preference to D.

asciilifeform: thereby any peer ranking algo that favours 'useful' peers, will give A,B,C a substantial advantage over any new D.

asciilifeform: gedankenexperiment : let's suppose that all new blox presently are born in nodes A, B, C. from there, propagate to rest of planet. the immediate peers of A, B, C, will have them parked at the very top of their peer table, right below the manually-added 'meat' peers. now say a D shows up. D may have faster hash rate even than A+B+C, but his blox will still be considered only ~after~ those of A,B,C. at least, for a time.

asciilifeform: i'd be interested to see what mircea_popescu makes of this algo. because i can see a possible fatal downside.

asciilifeform: imho a trb node ought to only advertise addrs that either a) part of the manual --addnode set the node was brought up with , or b) actually supplied a correct block in the recent past

asciilifeform: mod6: my main hesitation re the item is that it may be 'plugging wrong end of funnel', quite possibly no such gymnastics would be necessary if we simply had a http://btcbase.org/log/2018-09-23#1853143 and always --addnode'd the actually useful nodes, rather than relying on pure chance ☝︎☟︎

asciilifeform: e receiver’s reassembly timer. For higher delay high capacity network paths, this limit of 65,535 packets in flight can be a potential performance bottleneck [RFC 4963].'

asciilifeform: in particular, 'Lost fragments represent a slightly more involved problem than lost packets. The receiver has a packet reassembly timer upon the receipt of the first fragment, and will continue to hold this reassembly state for the reassembly time. The reassembly timer is a factor in the maximal count of packets in flight, as the packet identifier cannot be recycled within the period defined by the sender-received path delay, plus th

asciilifeform: the other item of interest, not mentioned, is that quite a few infrastructural routers will simply drop frags, for the reason that all but 1 frag of a fragged packet contain no copy of the source or destination port -- and thereby impossible to ddos-filter. so they drop'em.

asciilifeform: diana_coman: i.e. there's a tiny reassembly buffer, and if it is occupied while new frags fall in, it drops

asciilifeform: diana_coman: do you have a list of the ones that reached the other end ?

asciilifeform: it's a network problem, not a trb one.

asciilifeform: trinque: not long ago noticed a 400 blox gap on mine, where interestingly restarting did not fix, had to sit for the typical n days and finally meet noads who actually have new blox and dun get malleus'd

asciilifeform: and from this, goes to have power of attourney and trademark rights? bit of a climb eh

asciilifeform: i still dun fully grasp the process whereby a theo 'marries' a bobbeck. what's the draw, exactly