log☇︎
74700+ entries in 0.486s
asciilifeform: if there is literally no way to memorialize the fact that i took $item off a dead nazi, or dug up from sunken atlantis, rather than wrote it personally -- said item will have to be done without. (unless someone else in your wot does the imho monumentally idiotic act of 'adopting it as own child') ☟︎
mircea_popescu: nevertheless, i am saying, flatly, that this isn't a knowledge. as per lieberman.
asciilifeform: slave-fucking was a sometimes-thing there, evidently.
asciilifeform: this is a consistend position.
mircea_popescu: children are not a goal.
asciilifeform: imho this is a distinct item and oughta remain indicable.
mircea_popescu: this is not a matter of "style", it is a matter of identity.
mircea_popescu: vtron that presses a tree containing more than one genesis is, by that very fact, a broken implementation.
asciilifeform: and i entirely agree with mircea_popescu that working with such a thing is terrifyingly poor discipline.
asciilifeform: every node in a tree with 1,001 roots, flows back to ~one~ particular root.
asciilifeform: you NEVER press 'a multiroot tree.' you press ~one node~, which has ~one root~
mircea_popescu: vtron that presses a multi-root tree is, by that very fact, a broken implementation.
mircea_popescu: i mean, it is a 1=2.
asciilifeform: this is what i meant by 'logically valid.' it isn't a '1 == 2'
asciilifeform: you pick a node to press to.
mircea_popescu: the concept of "a press" towards more than one root is actually devoid of content.
asciilifeform: it is a logically valid, if stylistically undesirable, state
asciilifeform: mircea_popescu: i disagree that multiple-roots should be a mechanical eggog.
asciilifeform: cultivating a project with multiple roots may be bad form, but vtron ~must~ display properly
asciilifeform: mod6: if your vtron throws out a valid, signed-by-wot patch that has 0 antecedents, it is broken
mod6: the main reason that I bring this up; 'foobar.vpatch' is being dumped out of the flow and causing problems since i've implemented the axiom of 'a vpatch can only be in the flow if all of its antecedents are present.' which causes a problem in this case as it gets chucked out as an orphan.
asciilifeform: mircea_popescu: recall, i actually had a wholly separate shiva-bridge
mircea_popescu: asciilifeform an item can exist in two copies : as its independent tree AND ALSO as a patch imported in a project.
mircea_popescu: but an important point here i must stress is the COGNITIVE LOAD of a v tree. these aren't mechanical "oh, it changed x file so it goes with x". you think, as the author, and exactly in the manner of scoring wot members : what should this item be anchored on ?
asciilifeform: setting up a wholly artificial dependence also rubbed me wrong
mod6: so say we go with that... that 'foobar.vpatch' just inherits an antecedent 'genesis.vpatch'; how does one distinguish between 'foobar.vpatch' as a root and 'genesis.vpatch' as a root without additional criteria?
mircea_popescu: all patches to trb tree must refer to an antecedent in the trb tree. no exceptions. if "nothing comes to mind", genesis is a fine default.
asciilifeform: at any rate 'solitary node' is a perfectly valid state for a vtree.
mircea_popescu: the latter dun exactly make it a leaf tho
mod6: <+asciilifeform> mod6: wtf means 'root and leaf at the same time' << in the old code, it is a root, because it has no antecedents, and it is a leaf because it also has no descendants.
mod6: ok i guess that would be bad in the sense that "i'm just staring this project, my genesis should be a root even though nothing descends it yet..."
asciilifeform: it was a perfectly valid vtree.
mod6: ok, so that's basically what I was wondering. to be designated a proper 'root', do we also add the criteria of "must have an actual proper decendant"
mircea_popescu: rule is very simple : all items with false inputs create a project. whole, independent.
mod6: so in the case of my previous implementation (V99995), these island-roots would simply endup at the end of the flow, as a leaf. as they are both a root and a leaf at the same time.
mircea_popescu: mod6 such a "patch" may not exist.
mod6: so this one single vpatch has a 'false' input, and a non-zero sha512 output.
asciilifeform: it's a second root and mircea_popescu really, really hates'em.
mod6: Say that we have a flow, like trb, with all sorts of vpatches that stem from one single root. where that root is designated as a 'root' because all of its inputs are 'false'. what should happen with a vpatch that, for instance, just adds a file to the source and has no antecedents, nor decendants.
mod6: So, my V changes that implement a proper wot-variant V are looking pretty decent. however, I thought up an edge case that I'd like to discuss a bit more -- and I know we've been over this a bunch, and even recently.
asciilifeform: 'Morales had no rope. And no fingers. He had a ten-foot length of bandage. And brass balls. Somehow, this dude with no fingers lowers himself from the window on the bandage. It snaps. Morales falls 20 feet onto an air conditioner, then another 20 feet to the ground. Injured, but alive. The Family and FALN whisk him away.' << first-class.
asciilifeform: theoretically there is a borderland of folx who straddle the edge of sanity
asciilifeform: iirc there was a spiffy mircea_popescu article about the costs of letting 'harmless' idiots live (makes idiocy seem acceptable)
mircea_popescu: takes me all of half an hour to organize a 10% chargeback pile, which is the immediate end, for isntance.
mircea_popescu: broken shit is what you get when you don't let the "overly agresive Alphas" beat shithead manning into a bloody pulp.
a111: Logged on 2017-01-25 22:32 asciilifeform: broken searches are nearly always a result of some pathetic piece of shit trying to 'be clever'.
a111: Logged on 2017-01-25 22:22 ben_vulpes: WHAT i ask you IS THE POINT of an ultra-heavy IDE that a) cannot tell me where the protocol breakage is during a refactor much less what to do about it b) has a project search SO MISERABLY BAD that a man must revert to grep in order to find ANYTHING
a111: Logged on 2017-01-25 21:47 phf: so if you have a system that you implemented fast, but it's slow, but you know how to now slowly make it fast, you have a strategy. if you're chasing corner cases, running a profiler and get mostly flat distribution, writing in special cases, etc. you don't have one
a111: Logged on 2017-01-25 20:59 asciilifeform: 'Why haven't you embraced Bitcoin to get away from the restrictions of the banks / credit card companies? - We used to accept bitcoins through Coinbase. They dropped us a year ago because we are a kinky site. No joke.'
mircea_popescu: ben_vulpes first mentioned (ever! in history!) here : http://trilema.com/2017/my-ai-problems-a-humble-confession/#comment-121007 then referenced in http://trilema.com/2017/dupa-dealuri/#comment-121112 and http://trilema.com/2017/how-to-deal-with-little-girls/#comment-121192
ben_vulpes: granted, it serves a different problem, "searching documents" with naive user input.
asciilifeform: (and the occasional literate d00d, e.g., phf, naively making use of a library written by 'clever' peopel)
asciilifeform: broken searches are nearly always a result of some pathetic piece of shit trying to 'be clever'. ☟︎
asciilifeform: esp. if you are already lifting a standard unix userland
asciilifeform: ben_vulpes: i find broken ~search~ particularly infuriating, because it is so easy to make a working (i.e. grep-compatible) one
ben_vulpes: WHAT i ask you IS THE POINT of an ultra-heavy IDE that a) cannot tell me where the protocol breakage is during a refactor much less what to do about it b) has a project search SO MISERABLY BAD that a man must revert to grep in order to find ANYTHING ☟︎
a111: Logged on 2017-01-25 21:44 phf: http://btcbase.org/log/2017-01-24#1606341 << optimization strategy is when you have a slow algorithm somewhere in your architecture (you put it there because it was a reasonable tradeoff at the point), but you can replace it with a faster algorithm without leaving too much damage on the architecture (alf calls it "scaring")
trinque: ben_vulpes: http://trilema.com/2017/my-ai-problems-a-humble-confession/#comment-121007
phf: so if you have a system that you implemented fast, but it's slow, but you know how to now slowly make it fast, you have a strategy. if you're chasing corner cases, running a profiler and get mostly flat distribution, writing in special cases, etc. you don't have one ☟︎
phf: http://btcbase.org/log/2017-01-24#1606341 << optimization strategy is when you have a slow algorithm somewhere in your architecture (you put it there because it was a reasonable tradeoff at the point), but you can replace it with a faster algorithm without leaving too much damage on the architecture (alf calls it "scaring") ☝︎☟︎
asciilifeform: 'Why haven't you embraced Bitcoin to get away from the restrictions of the banks / credit card companies? - We used to accept bitcoins through Coinbase. They dropped us a year ago because we are a kinky site. No joke.' ☟︎
asciilifeform: 'Let's say we move our servers to the newly formed country of Fetopia. The US, UK, Australia, Germany, etc. can still make it difficult for people within their countries' borders to access FetLife. I don't think we would do the community a favor by making FetLife harder to access.'
ben_vulpes: inb4 Framedragger builds a bitcoin kinknet
asciilifeform: unrelatedly, if phf or anybody else has a copy of mulisp-90, plz to post. ☟︎
mircea_popescu: refused to use a condom!
mircea_popescu recalls a time in the 90s... concidentally at the time the world was last great i was too young to care much about women.
a111: Logged on 2014-09-17 02:47 asciilifeform: http://www.tpnn.com/2014/08/07/video-russian-laser-show-features-obama-sucking-on-a-banana
mircea_popescu: i feel this will make a lot of difference.
asciilifeform: in other lulz, http://archive.is/j41JR >> 'A group of at least seven protesters, apparently with Greenpeace, are scaling a crane in downtown Washington DC in protest at US President Donald Trump. ... the protesters unfurled a 35ft by 75ft banner reading “Resist,” 270ft above the ground. '
asciilifeform: the lulzy bit is that soft-warehouse (makers of subj, and one-time mega-monopoly, they made the only good lisp for msdos pc) was bought by ti and put in a hole, and cement poured.
asciilifeform: changed by adding other words to `Derive', nor by your belief that the word `derive' is an accurate description, using the correct terminology, of the purpose of the software. In pursuit of your academic research project, please rename your software to avoid using `Derive' or any other word that might cause a reasonable likelihood of confusion with `Derive''
asciilifeform: 'A failure in the data communications system of the US-made weapon system has been blamed. Lockheed Martin, based in Maryland, builds and maintains the missiles.'
asciilifeform: unrelatedly: in re the metal lisp thread: http://picolisp.com/wiki/?PilOS << i could not get this to boot. but it is ~exactly what i had in mind (i'd much prefer that it be with a scheme, rather than a crackpot dynamically-scoped 1970s lisp, but the idea is illustrative)
asciilifeform: iirc we had a thread here about ritual concessions to thieves, folx placing car keys where thief will find'em, etc
asciilifeform: cheaper to replace a glass than to drywall a new wall.
asciilifeform: i always suspected that the glass doors are a kind of 'fuse'
asciilifeform: does it matter what kind of door, on a paper house /
mircea_popescu: phf nah, the idea is you get exactly like current, "security door" except the half inch steel sheet is on the inside, then 3 claymores stacked, then a thin al profile on the outside facing. have a shock detector modeled on 90% of the kinetic delivery of the standard issue us invasion corps battering ram.
phf: hmm, i can't recall anyone doing it in the 90s. i think this is a uniquely american problem though, where you can't trust frame to hold the door in place. traditional solution is to simply put the door on welded frame inside a concrete hole, so you will need something better than a claymore to dislodge it
mircea_popescu: there's a huge space there to pick a threshold.
mircea_popescu: ~nothing easier in the world than making a team of us experts shit their pants and spend the rest of the day loitering waiting for "the conversation" to happen.
mircea_popescu: and you're wrong, a claymore going off in response to a hammering of the door will postpone the raid by HOURS
asciilifeform: which can indeed be a difference of 3 vs 1 sec.
asciilifeform: the key zap in a fighter jet
mircea_popescu: my thought was more along the lines of "if the difference between a second and a minute has a practical effect, there are other broken parts of the whole process that need fixing"
asciilifeform: you want a zap that cannot be cancelled, even by mechanically separating the 'zapper' from the payload with explosive.
asciilifeform: ideal eraser is probably a 'belt and suspenders' approach, where the wire is pulled through neodymium magnet (with hand crank backup) but meanwhile a pancake magnet is lowered onto the spool (for rapid, though imperfect, erase, in case there is not time to wind the entire tape) and at the same time current is run through the steel wire, if a current source is available.
mircea_popescu: maybe we should have a teat pulling sit-in over this.
mircea_popescu: "oh, if the chinese do qke, the us army will... have to find a way to diddle all keys in subatomic transit"
asciilifeform: even the humble steel wire recorder, ancestor of magnetophone, could easily come back as a practical thing.
mircea_popescu: asciilifeform i'm not discussing practical utility, merely that it makes sense from a technical perspective. unlike the other thing.
asciilifeform: btw magnetic tape really shines in otp-land. it costs a penny a GB, and you can trivially write two tapes identically with same head (plugged into rng naturally), AND -- bonus -- can destroy it immediately when it comes out of read head.
mircea_popescu: anyway, re qc, bonus lulz for the huy-guy : so the chinese are trying to do it over satellites, because currently it's been demonstrated over much shorter distances. and the dod bright lightbulb (seriously, he is probably the smartest guy there, seeing how he has a blog) went "oh, this would allow safe key transfer, well tell you what, the army could DoS the system!"
asciilifeform: mircea_popescu: 'quantum communication' is golden toilet nonsense, for the cost of building the link, you can perma-sync a thousand years of OTP.
asciilifeform: as for 'how to even know whether it works', that one's easy. qc is strictly a wunderwaffen for breaking rsa. if it breaks rsa -- it worx. otherwise, not.
mircea_popescu: there isn't even a measuring tool ; we have nfi if any observed effect is even anything.
mircea_popescu: we even have a multitude of strong statements of impossibility all converging towards a hearty "fuck you" located at the core of this concept
mircea_popescu: asciilifeform i hold they are not, for the following reason : the nonsense of contained cold fusion rests on mechanical problems, ie, the plasma expands and therefore cools too fast. at least we have SOME TOOLS to do mechanical work with. like you know, a fulcrum. perhaps the plasma could be put in a fulcrum somehow ? or maybe with cogs ? dunno, should at least investigate.
mircea_popescu: literally, mana from heaven is not a more contortedly i-read-penthouse-in-physics-class
asciilifeform: it's a screamingly single-purpose (hypothetical) mechanism.