log☇︎
445600+ entries in 0.279s
mircea_popescu: it receives alt block 6, timestamp 990. it rejects it
mircea_popescu: it receives block 6, with timestamp 1010. accepts it.
mircea_popescu: at real timestamp 5000 pogo starts
mircea_popescu: so, pogo has on disk last block 5, bvlock timestamp 1000
mircea_popescu: lemme model this maybe i'm saying it stupidly.
ascii_field: in order for it to be rejected. the chain on disk will have to be no more than a day behind
mircea_popescu: doesn't matter how old that block is. the block coming after it will be ~10 minutes later.
mircea_popescu: because when hitler block shows up, the disk has a block it ends with
ascii_field: how old is the chain on disk
ascii_field: how does node know this?
ascii_field: then genuine block appears. it now appears to be 'from the future'
ascii_field: next block is from hitler. it has timestamp of... yesterday.
mircea_popescu: jurov 5 couldn't have been that far off and be accepted from 4.
mircea_popescu: now, this theoretically is vulnerable. but practically, can you construct the attack for me ?
jurov: because you mined block 5, with a bad time, someone else mines 6 << and it will get dropped because pogo considers delta from 5 invalid
mircea_popescu: once it accepts a further block, it resets its clock too.
mircea_popescu: suppose it's not empty. so, pogo thinks time = last block on disk
mircea_popescu: can you forget that for a moment ?
ascii_field: how did the disk become not empty ?
mircea_popescu: let's consider the not clean disk situation
mircea_popescu: forget that for a moment.
ascii_field: all the malicious miner needs to do is keep nudging withing the allowed bounds
mircea_popescu: no average. set clock = timestamp of last block
ascii_field: there is no 'bad time', normally there is 2hrs of 'play' in the gears
ascii_field: all it sees is 'block was broadcast, has timestamp t, average it now'
mircea_popescu: because you mined block 5, with a bad time, someone else mines 6 with a right time, pogo is now on block 6.
mircea_popescu: this is a horribru example. so, miner has 10% of the hash, 10% of the bnlocks are badly timed. what of it, pogo resets on the remainder.
mircea_popescu: what % hash does this miner have ?
ascii_field: fine, example. malicious miner starts crapping out blocks with timestamp more and more off in the future
ascii_field: you are still open to crafted, induced clock drift
ascii_field: even if you weasel out of this one by hardcoding block hashes, etc
ascii_field: much less what time
ascii_field: because the box has no idea what year it actually is
ascii_field: mircea_popescu: if we lack a clock, we are wide open to replay attack
ascii_field: realize, time is specified as an invariant
mircea_popescu: again : if you have block x, from a year ago, when you get block x+1, the styamp on block x+1 will be, 1year ago + 10 minutes
ascii_field: according to the traditional bitcoin protocol.
ascii_field: >2h --> too far
mircea_popescu: the next block to wahtever it has on disk is not going to be too much from the future.
ascii_field: and any incoming block is invalid, because 'from the future'
ascii_field: from the cpu's pov
ascii_field: and if something on disk, then perhaps it is 2009
ascii_field: because machine thinks it is '69
ascii_field: it will be 'from the future'
ascii_field: it'll be 2 hrs delta to the last known!
mircea_popescu: but the next block it gets wouldn't be more than two hours off.
ascii_field: for one thing, there may not be any blocks on disk
mircea_popescu: i have. but tell me again.
ascii_field: see today's thread
mircea_popescu: and setting the time to "timestamp of the last block it actually has on disk" is going to what ?
ascii_field: but will agree to plug in box
ascii_field: because these are to be given to folks who will not volunteer any meat commitment
mircea_popescu: and you don't like having to start it by manual command because myeah.
ascii_field: this doesn't need any magic, it works on trad kernel
mircea_popescu: i mean, by siomething running on the box in question
ascii_field: i said this
ascii_field: it keeps time fine!
ascii_field: mircea_popescu: yes, though i've forgotten the exact number. 500MHz i think it was.
ascii_field: the idea, as i understood it, was to let folks set up nodes without human committment
mircea_popescu: ascii_field is the things' processor freq known ?
ascii_field: they never get off the ground at all
ascii_field: with the scheme described, you don't even need an attack
danielpbarron: isn't there supposed to be a way to send commands to the pogo? otherwise how is it useful to me as a "full node" if I can't query it for blocks/transactions and relay new ones I have created for it?
mircea_popescu: held together by the superficial tension of hemorhagic diarhea
ascii_field: i am not sure how this improves on anything
mircea_popescu: nevertyheless, that is a marked improvement over the current situation
mircea_popescu: it is held together by the superficial tension of frog entails.
ascii_field: not 'learn to maintain this steam engine'
ascii_field: whole point was 'here let me fuck you and plug this box into your router'
ascii_field: what, i'm to poll it ?
ascii_field: congrats, now pogo behaves like the clock on my stove
mircea_popescu: you take it down and start it over with new magic touch
mircea_popescu: yes this opens it up to attack, but
ascii_field: howabout at 4am on a sunday when the mains flickers ?
mircea_popescu: from there on, iut carries on by simply syncing its clock to last accepted block.
mircea_popescu: this, you have to do. like it or not we dunno how to abstract it away.
mircea_popescu: so, when you start the pogo, you provide it with the human touch, divine spark, politica time.
mircea_popescu: anyway, re the switch : let's talk it through.
assbot: Drone fires burgers at the homeless in charity viral video gone wrong - Americas - World - The Independent ... ( http://bit.ly/1HdQJcW )
mircea_popescu: oh, openssl put out a bug ? best make sure the fucking calculator pulls it in.
mircea_popescu: ascii_field i know i'm fed up with random shit depending on random shit, but it's not mere stupidity. it's just them trying to make sure shit permeates everywhere.
ascii_field: everyone who ever made fun of rms deserves to be fed a litre of 'toe jam' now.
ascii_field: at this point, any unixlike box with a graphical display may as well be running winblowz
trinque: nowhere clearer than the recent 1m+ lines-changed linux release
ascii_field: .... and how much shitgnomism is rolled into the 'updated' version of 10,001 proggies that will install.
trinque: I've half considered trying to stop using the web so much, aside from maybe syncing certain sites to a local box.
ascii_field: and that's ~after~ resolving the circular dependencies
ascii_field: trinque: and guess what, rebuilding all the deps of that thing would take a month+ on my box
trinque: now that the de-dockerization is complete, I may have it announce its current balance along with newly uploaded deeds
trinque: should be good for a while though
trinque: and if it gets low, feel free to throw some bitdust in the tank
trinque: it will switch to the next
trinque: lol and chrome has it too
trinque: ascii_field: did you see the awesome glsa on icu yesterday?
shinohai: Hmm missed that. Interesting concept
assbot: Logged on 07-07-2015 04:08:49; decimation: asciilifeform: https://github.com/zrm/snow < I thought this could be it, but then saw the deps list
assbot: Need a search term.
trinque: except that C++ itself foils any attempt
trinque: I cannot see how this tool does not exist
ascii_field: trinque: the 'dot' file, does it contain something like a logical flow graph ?