asciilifeform: mircea_popescu: tldr - asciilifeform failed to square the circle
asciilifeform: ( or, what, some derp mining on a 486, can cause space inside a block to be filled ?! )
asciilifeform: danielpbarron: and i can hardly picture the low-density 'subcoinbases' actually ever making it into a block .
asciilifeform: where you have some ~reason~ to prefer high-density objects
asciilifeform: as i see it, what is missing here is the knapsack per se
asciilifeform: i.e. what does the vectorization actually accomplish ?
asciilifeform: how is this different from the old-fashioned scalar coin where the ~quantity~ decreases ?
asciilifeform: (and, more importantly, how does a system having coins taking the form (a,b), rather than scalar (c), not reduce to same (c) == a*b ? )
asciilifeform: what does one do with 'low-quality coins' ?
asciilifeform: is it the knapsack problem, where coins are now vectors, rather than scalars, they have a volume and a density ?
asciilifeform: danielpbarron: educate me ( a non-euloran ) what means 'quality of coin'
asciilifeform: and (as iirc mircea_popescu observed also) it would reduce to ~same situation as today, where the folx with the serious hashing iron would eat ~all of the cake, and everybody else gets to do the laundry.
asciilifeform: tx-sized blocks, or if you like, block-sized tx..
asciilifeform: danielpbarron: it does seem to reduce to the earlier 'let's say mining were what you did to make a tx, and there were no blocks as such' neh ?
asciilifeform: danielpbarron: didja write this up at length anywhere ?
asciilifeform: ( now! for all i know, they wrote directly to mircea_popescu . but notice that he is not burning with the desire to share this fact. the secrecy incentive remains even for folks N degrees separated from a known miner ! )
asciilifeform: even though such high rollers as mircea_popescu , sometimes transmit tx through it.
asciilifeform: observe, no miners have written to asciilifeform asking for ssh-wire to dulap.
asciilifeform: afaik the real boojum is that miner has overwhelming incentive to stay secret, and the more powerful -- the moar so
asciilifeform: in so far as i can unmistakeably ~determine~ that my tx was not mined -- and know when it ~is~ -- i can transmit. it can take potentially infinite time...
asciilifeform: but i can transmit a tx. (for the time being)
asciilifeform: where you have 6 miners and their 6 nodes, being 'the network'.
asciilifeform: the other fundamental problem is that classical bitcoin comes with immense incentive for miner cartelization. if nothing were changed other than 'miner must be a proper node', we get what amounts to visa.
asciilifeform: currently it is possible to break even as a miner without accepting ANY tx
asciilifeform: trinque: aha. the running thread was re: how might they be made to
asciilifeform: they would actually rather mine coinbasetx-only blocks
asciilifeform: because miners have more incentive for secrecy than they have for gathering txes from the wild.
asciilifeform: but in fact if this becomes common -- and you can think of the spamola attacks of last 2 yrs as in fact attempts to MAKE it happen -- propagation will stall.
asciilifeform: atm mempool works as a 'meat market' where the eligible chixx stand around, waiting, hoping for a serious mircea_popescu to show up and take'em home
asciilifeform: gotta specify why the hypothetical conditions will differ from the current ones, trinque
asciilifeform: afaik nobody in tmsr has any direct link to any miner whasoever (or at the very least, wants to admit to it)
asciilifeform: trinque: say trinque wants to transmit a tx
asciilifeform: i could picture some clever mathemagical route whereby each hop can only take a portion of what the n-1-th node consented to -- but i know of no algo to make it thinkable.
asciilifeform: (even if you limit the total node feed to some small constant, the miner can ~still~ take ~100% of it, this way)
asciilifeform: if you let ANYONE, under ANY circumstances, appropriate some of the value of a tx without the consent of its original author, you create a sybil-feeder, where the last hop (i.e. the miner) can simply eat 100% by simulating the passage of the tx through 1,000,001 hops of fictional nodes.☟︎
asciilifeform: oooook i finally realized that the problem -- as stated above -- is unsolvable
asciilifeform: also if 'a block has many fathers', as in contemplated scheme, this re-introduces the possibility of pool. which imho is a Bad Thing.
asciilifeform: ( the above ^ now that i think about it, could be simplified to mircea_popescu's earlier 'node accepts if you put an output to his addr in your tx' )
asciilifeform: and this kind of scheme would also nuke 'canned tx', as discussed earlier. so not really such hot stuff.
asciilifeform: i can think of ~one~ approach, so far: tx creator asks his first-hop node for a nonce, which he then incorporates into his tx, which protocolically declares consent to the node fee. similarly to how miner fee already works. BUT this does not solve the problem of how 2nd ... nth hops, could add anything whatsoever meaningful to the tx.
asciilifeform: this may very well resolve to mircea_popescu's unsolved 'blind inputs' problem.
asciilifeform: but at the same time has to preserve the validity of the original tx creator's signature.
asciilifeform: (it has to be ~protocolic~, that is, something that the next-pass relay, or miner, could not simply strip out)
asciilifeform: and now the interesting observation -- what's with the plastic parts (fan blades, connector headers) that did not melt or so much as warp ?
asciilifeform: in other lulz, 'medium' now replaces images with blank turds whenever archive.is (and also ye olde archive.org !) saves a page there
asciilifeform: (before anyone laughs, i will point out, yes, the necessary mechanical parts for this do not presently exist.)
asciilifeform: !~later tell mircea_popescu here's another crackpottery in re the nodes/miners/txers 'racul, broasca si o stiuca' : ~multistage mining~. where a node can encumber (protocolically/mathematically - for now i will not specify how) a tx with some proofofwork, when passing it on to next relay; and when the tx is mined, the block reward is split between the multiple parents of the final tx.
asciilifeform: damn, i was hopin' he'd tell us about something he'd... gone, and gotten
asciilifeform: just like 'miners are nodes are txers' was jettisoned in, what, 1st year of bitcoin
asciilifeform: if 'spec' contains an 'octopus gland of death', item that can be safely and unilaterally jettisoned by a player, it'll be jettisoned
asciilifeform: this is an ad-hoc, orcish version of the 'pay to play' discussed earlier.
asciilifeform: which is not connected to mempool in the usual sense (why would, e.g., antpool, want to tell other pools your tx)
asciilifeform: then it presumably goes into the pipe.
asciilifeform: at this point it is not a mega-seekrit, that mempool only barely worx.
asciilifeform: but it does not solve the 'nodes women' thing.
asciilifeform: this is not a strike against mircea_popescu's algo -- imho it is direly necessary
asciilifeform: (given mircea_popescu's algo, they more or less must vertically integrate.) then there will be equally little point for nonmining nodes to operate as there is today.
asciilifeform: incidentally mircea_popescu's algo doesn't force miners to ~relay~, necessarily, blocks
asciilifeform: the other problem is that 'tx is paid for all nodes for a million blocks, while creator pays once and miner rakes in the cake' is not a substantial improvement over the current case
