Hide Idle (>14 d.) Chans


← 2020-08-04 | 2020-08-06 →
verisimilitude: http://logs.nosuchlabs.com/log/asciilifeform/2020-08-04#1017833 I wholeheartedly agree, asciilifeform. I've found it queer that programs are written, with changes measured line-by-line, in such a jilted fashion, where the popular debugging method is to revert changes that break, rather than seek true understanding.
snsabot: Logged on 2020-08-04 22:06:01 asciilifeform: the problem isn't 'outsource', but the very idea that hundred+ people can 'write software together', is crock of shit
verisimilitude: http://logs.nosuchlabs.com/log/asciilifeform/2020-08-04#1017850 The main issue with timestamps, in my eyes, is not necessarily the odd timekeeping methods in common use, but the fact that time is an unbounded quantity, so using a fixed-size to handle it is fundamentally wrong. Discarding with time is an automatic simplification.
snsabot: Logged on 2020-08-04 22:12:10 asciilifeform: timestamps are intrinsically bogus, in my lights.
gregorynyssa: http://logs.nosuchlabs.com/log/asciilifeform/2020-08-05#1017864 << I call this "evidence-based programming" and it is indeed the dominant paradigm within our present industry.
snsabot: Logged on 2020-08-05 01:20:44 verisimilitude: http://logs.nosuchlabs.com/log/asciilifeform/2020-08-04#1017833 I wholeheartedly agree, asciilifeform. I've found it queer that programs are written, with changes measured line-by-line, in such a jilted fashion, where the popular debugging method is to revert changes that break, rather than seek true understanding.
gregorynyssa: the ideas of XP, "scrum," and "agile" are basically outgrowths of this dismal worldview.
trinque: themselves outgrowths of yes, cheap, ever-printed money.
trinque: "run as fast as you can just to stay where you are" if that.
asciilifeform: http://logs.nosuchlabs.com/log/asciilifeform/2020-08-05#1017866 << note however -- a 128-bit epochtime runs long past sun burning out.
snsabot: Logged on 2020-08-05 01:22:34 verisimilitude: http://logs.nosuchlabs.com/log/asciilifeform/2020-08-04#1017850 The main issue with timestamps, in my eyes, is not necessarily the odd timekeeping methods in common use, but the fact that time is an unbounded quantity, so using a fixed-size to handle it is fundamentally wrong. Discarding with time is an automatic simplification.
asciilifeform: ( afaik 128b epoch not in use anywhere, because idiots )
verisimilitude: A one hundred and twenty-eight bit integer is also usually large enough, but that doesn't discredit arbitrary-length arithmetic. The point stands; eliminating time is a better solution.
asciilifeform: verisimilitude: the trouble w/ arbitrary-width arithmetic is that there's 9000 places where yer stuck with a finite bitness. e.g. packets.
verisimilitude: Timestamps have the same issue.
asciilifeform: !w poll
watchglass: Polling 12 nodes...
watchglass: 205.134.172.4:8333 : (172-4.core.ai.net) Alive: (0.033s) V=70001 (/therealbitcoin.org:0.7.0.1/) Jumpers=0x1 (TRB-Compat.) Blocks=642360
watchglass: 205.134.172.26:8333 : Alive: (0.091s) V=99999 (/therealbitcoin.org:0.9.99.99/) Jumpers=0x1 (TRB-Compat.) Blocks=642360
watchglass: 205.134.172.27:8333 : Alive: (0.105s) V=99999 (/therealbitcoin.org:0.9.99.99/) Jumpers=0x1 (TRB-Compat.) Blocks=642360 (Operator: asciilifeform)
watchglass: 205.134.172.6:8333 : (172-6.core.ai.net) Alive: (0.146s) V=99999 (/therealbitcoin.org:0.9.99.99/) Jumpers=0x1 (TRB-Compat.) Blocks=642360
watchglass: 108.31.170.3:8333 : (pool-108-31-170-3.washdc.fios.verizon.net) Alive: (0.225s) V=99999 (/therealbitcoin.org:0.9.99.99/) Jumpers=0x1 (TRB-Compat.) Blocks=642360 (Operator: asciilifeform)
watchglass: 208.94.240.42:8333 : Alive: (0.190s) V=99999 (/therealbitcoin.org:0.9.99.99/) Jumpers=0x1 (TRB-Compat.) Blocks=642360
watchglass: 192.151.158.26:8333 : Alive: (0.206s) V=70001 (/therealbitcoin.org:0.7.0.1/) Jumpers=0x1 (TRB-Compat.) Blocks=642360
watchglass: 143.202.160.10:8333 : Alive: (0.171s) V=70001 (/therealbitcoin.org:0.7.0.1/) Jumpers=0x1 (TRB-Compat.) Blocks=642360
watchglass: 213.109.238.156:8333 : Alive: (0.335s) V=99999 (/therealbitcoin.org:0.9.99.99/) Jumpers=0x1 (TRB-Compat.) Blocks=642360
watchglass: 176.9.59.199:8333 : (static.199.59.9.176.clients.your-server.de) Alive: (0.288s) V=99999 (/therealbitcoin.org:0.9.99.99/) Jumpers=0x1 (TRB-Compat.) Blocks=291039 (Operator: jurov)
watchglass: 188.121.168.69:8333 : (rev-188-121-168-69.radiolan.sk) Alive: (0.347s) V=99999 (/therealbitcoin.org:0.9.99.99/) Jumpers=0x1 (TRB-Compat.) Blocks=642360
watchglass: 103.36.92.112:8333 : (terebe.ns01.net) Alive: (0.643s) V=99999 (/therealbitcoin.org:0.9.99.99/) Jumpers=0x1 (TRB-Compat.) Blocks=642360
asciilifeform: verisimilitude: in principle, a 256-bit timestamp would span the entire life of universe, as per supposed state of art, from start to hypothesized end. in nsec.
asciilifeform: ( in practice, prolly unnecessary, e.g. the 8k years that span civilization of egypt, fit in < 38bit seconds ... )
verisimilitude: It also must be noted that it's a waste of space. My preferred method, must I've timestamps, is to choose a larger unit. Second-granularity on, say, server access logs, is perhaps unnecessary in some situations, and granularity as large as an entire day would be reasonable. A twent-yfour bit day count would last tens of thousands of years, which is ``good enough'', if it must be recorded.
asciilifeform: verisimilitude: on the contrary, i often find that second is insufficiently granular for logs, where a process can piss out 9000ln/sec., for guaranteed preservation of ordering
verisimilitude: It depends on the server, I know.
asciilifeform: depends naturally on what yer loggin
verisimilitude: For an HTTP or Gopher server, I wouldn't really care for anything more than a few hours accuracy, and proper ordering within the hours.
verisimilitude: A timestamp acknowledgement server, for reasonably guaranteeing the existence of a file at a certain time, would likely be fine with second-granularity.
verisimilitude: It depends on who the users would be.
asciilifeform: verisimilitude: current state of art for 'guaranteed time', deedbot, gives multi-hour granularity (i.e. until block is mined)
verisimilitude: I'd in mind, in particular, the SPARK program mentioned in chapter eight of that book I reviewed.
asciilifeform doesn't see much point in washington-style 'magic box in palace' centralized timestamping services
verisimilitude: It seems like it has an obvious use, in the proper context, there a university.
verisimilitude: I don't mind a magic box, so long as it's my magic box.
asciilifeform: right, clock is useful to owner. but anyone who relies on 3rd party clock as absolute authority, is simply a chump .
verisimilitude: I like how the linked bitcointalk forum discussion contains this: ``One or more bitcointalk.org users have reported that they strongly believe that the creator of this topic is a scammer.''
verisimilitude: Am I to understand not every topic is prefaced with this?
asciilifeform: verisimilitude: unlike many of the other ex-#t folx, asciilifeform never frequented that forum. but iirc it had various comically ad-hoc pseudo-WoT mechanisms, yes
verisimilitude: Oh, what it would change were I involved with bitcoin from the beginning; I probably wouldn't've had some of my nice ideas, however, so it evens out.
asciilifeform: revisiting clocks -- see also old thread re 'the key moving part of bitcoin is the decentralized monotonic timestamp', i.e. if had nuffin else but it, could trivially implement broad class of reliable mechanisms for multi-party agreement
snsabot: (trilema) 2017-03-02 asciilifeform: it is worth recalling the gedankenexperiment where it turns out that all you'd need to build 'martian bank' on earth, is if martians merely supplied an infallible 'martian clock', a service whereby they take a string S , at regular, say, daily, interval, and return its hash H
verisimilitude: So, bitcoin plays the ten-minute martian, then?
asciilifeform: verisimilitude: 100% of the mechanical pieces existed in '90s (chaum et all), the item that made present-day bitcoinism possible was specifically the longest-chain hash work mechanism
asciilifeform: ~10 avg., in practice varies wildly
verisimilitude: I recall a business put ads in the New York Times for a similar purpose.
asciilifeform: it's possibly world's least accurate electronic clock. its only virtues are monotonicity + immutability.
asciilifeform: verisimilitude: current-day nyt (and other fishwraps) routinely monkey w/ archives (and charge iirc significantly moar for ad that makes its way into printed birdcage liners, vs. www-only)
verisimilitude: Wouldn't the least accurate clock be a clock which is perfectly incorrect, and thus useful in how it can be trivially made correct by a constant transformation?
asciilifeform: verisimilitude: recall from grade school, 'accuracy' vs 'precision'
verisimilitude: That is this: A line which always bitflips is perfectly reliable.
asciilifeform: the figure on upper left corner -- 'accurate, but not precise' -- reflects btc picture of time.
verisimilitude: I stretched the thought too far, perhaps.
asciilifeform: verisimilitude: arguably i mislead in the earlier summary, oughta have said 'least precise'
verisimilitude: Alright. So, asciilifeform, I'm correcting a minor flaw I found in my MMC today, and will document it in my Finger service; what programming has been done lately?
asciilifeform: point being, knowing block # gives you a quite loose picture of the 'absolute time'. rather on the level of photo of night sky on given time.
asciilifeform: verisimilitude: after lengthy dry spell (heathendom matters) returned to cleaning up ch21b for publication.
asciilifeform: ( 21b is the derivation of the algo + proof ; 21c -- discussion of code itself. the code is written and polished, since early may )
verisimilitude: Oh, that reminds me; it won't be considered poor form if I leave a comment on every chapter as I read it, right? I could see the frustration of ``That's covered next chapter.''.
asciilifeform: verisimilitude: don't hesitate to comment
asciilifeform: verisimilitude: have you been reading for some time, or recently picked up ?
verisimilitude: Alright; I'll only hesitate to read; there's always so much reading to be done, although that's better than having none.
verisimilitude: I've commented on at least the first chapter.
verisimilitude: I certainly intend to finish it, know.
asciilifeform: verisimilitude: i wrote in , imho, sufficiently small chunks that can be eaten in <1h per ch.
asciilifeform: ( longer if you intend to attempt the puzzles, perhaps )
verisimilitude: It's a nice writeup, asciilifeform, and I'll certainly read each chapter as if my life depended on it, so that I may be the one who discovers any flaw to then report.
verisimilitude: Are earlier chapters updated with notes on flaws found?
asciilifeform: verisimilitude: i was mildly disappointed that none of the historical readers found the mine in ch15 . but i also did not find until later, so haven't any room to laugh at others
verisimilitude: I'll try my best to have an excuse to laugh at others.
asciilifeform: verisimilitude: the linked item thus far is the only lethal bug found & revealed.
asciilifeform: and yes i put a link in orig. ch15 text to the fix.
verisimilitude: Alright.
asciilifeform: 21 is taking as long as it is because specifically intend to rule out the ch15 situation by posting rigorous proof from beginning.
asciilifeform: ( a la ch14 )
asciilifeform: while on subj of programs that take yrs -- picked back up also my gnat-on-bare-x86 (orig. based on ave1's demo for linux)
asciilifeform: idea being, single-task msdos-style os, bootable from <1MB storage (incl. rom, if preferred) that contains peh, simple text editor, speaks via rs232 port (incl. to FG) .
asciilifeform: with zero heathen coad inside.
asciilifeform: notion being, can perform cryptographic operations purely in ram (w/out littering nonvolatile storage w/ intermediates) , on just about any old x86 junk box.
verisimilitude: That's certainly a neat idea.
verisimilitude: Without a schedule, it's far too easy to let goals slip by; from now, I'm going to aim for one chapter of FFA weekly, asciilifeform.
asciilifeform: it aint the logical conclusion of peh (that'd be a 100% clean iron on e.g. 'ice40') but has the advantage of ~0 cost takeup .
asciilifeform: verisimilitude: at one time i published ffa 1/ch/wk like clockwork. then at one pt no longer could (day job ate time) and this caused epic wank, began the process whereby broke up w/ mp et al
verisimilitude: Oh, I'm not too familiar about that drama.
asciilifeform: verisimilitude: historical interest strictly
verisimilitude: I recall ``asciilifeform did nothing'' which was clearly false.
verisimilitude: I should revise my SHA articles so they're more interesting to read, with hindsight; I've noticed next to nothing on loper-os is touched after publication, but several of my articles have been greatly improved by multiple rewrites, so it's worth considering, asciilifeform.
asciilifeform: if asciilifeform , like some of his critics, had been sitting on $millions, could've entirely wrapped up ffa in '17. but 'if wishes were horses' etc.
asciilifeform: verisimilitude: normally when i amend an item after publication, i leave a clear mark to indicate, with date.
asciilifeform: ( typos and similar excepted )
verisimilitude: I hold the idea of, since I could lie or forget about the timestamp, along with not caring, I never leave any.
asciilifeform: verisimilitude: i leave time marks largely for self, when reading yrs later, helps to have rough chronology
verisimilitude: I like subverting expectations, and anyone who thinks websites should have update notices should be made to recall that any of them could be lies.
asciilifeform: verisimilitude: sorta why i dun think much of timestamps in e.g. gpg sigs (and have no intention of including concept of time in peh)
verisimilitude: Not only can they lie, they can simply never be present, and I'm certain ``trustworthy'' sources gaslight in this fashion constantly.
asciilifeform: of course they do.
asciilifeform: verisimilitude: this is the essence of the concept of 'promisetronic'. in other words, the diff. b/w bitcoin and a heathen bank.
asciilifeform: or b/w gpg and heathen idiocies a la 'protonmail' & similar
asciilifeform: there's a whole ocean of 'homeopathic crypto' where 'secure' until&unless rando third party happens to decide to lie one day
asciilifeform: the 'ssl' tower o'shit is simply the best-known example.
asciilifeform: there's, sadly, no shortage of examples. say, the 'encryption' supposedly built into some brands of ssd .
asciilifeform: ( where user has 0 way of knowing what actually happens internally, whether the pw is retained in plaintext (in all units reversed to date -- it was in fact squirreled away in the controller somewhere ) etc.
feedbot: http://mvdstandard.net/2020/08/lebanon-puts-all-of-their-port-officials-under-house-arrest-after-massive-explosion-yesterday/ << The Montevideo Standard -- Lebanon Puts All Of Their Port Officials Under House Arrest After Massive Explosion Yesterday
newland0: to me, items that are published benefit from having a timestamp for reader convenience
newland0: 2020-08-05T22.53.27Z for example
newland0: a vpatch would be very unlikely to need anything outside of that range or precision
asciilifeform: newland0: nuffin stops anyone from placing timestamps in the text. in fact, custom for a while has been to place btc block #s in manifest.txt .
newland0: cool, that works
newland0: so what can i do with all of this, that might be incidentally of some benefit to others? stand up a node maybe?
newland0: how many GB does that require these days?
mats: ~316gb
asciilifeform: wb mats
← 2020-08-04 | 2020-08-06 →