00:13 |
* |
asciilifeform suspects that verisimilitude misinterprets the notion behind dunbar's number |
00:15 |
asciilifeform |
... it aint about 'typical human head can only remember N names'. tho may be fact. is rather about 'typical human has the meatware to give a rat's arse about N people'. (often enuff N , in modern degenerate anthill dweller, ~= 0) |
00:15 |
asciilifeform |
... but in no case N > coupla hund. |
00:15 |
verisimilitude |
I don't have to be some degenerate bugman to not care about others. |
00:16 |
verisimilitude |
I suppose. |
00:16 |
verisimilitude |
Now we reach what ``giving a rat's arse'' or ``caring'' means. |
00:17 |
asciilifeform |
variously interpreted, but generally 'see as playing character rather than npc' |
00:18 |
verisimilitude |
Alright. |
00:19 |
verisimilitude |
Sure, in other IRC channels, I certainly don't believe everyone else to be alive as I'm. |
00:19 |
verisimilitude |
So, a few dozen, sure. |
| |
~ 22 minutes ~ |
00:42 |
thehorrors |
[http://logs.bitdash.io/asciilifeform/2022-05-04#1099811] << I wonder if something like [http://kelp.sourceforge.net] could help with that |
00:42 |
bitbot |
Logged on 2022-05-04 12:19:02 asciilifeform: ... whereas if not embedded, gotta be coupled w/ the proggy ~somehow~ |
00:43 |
thehorrors |
On the other hand, I alway saw the extensive approach (e.g. test coverage) to be more transparent (e.g. sqlite autismo seems to be pretty effective) |
00:44 |
asciilifeform |
thehorrors: tests have 0 to do with subj |
00:44 |
dulapbot |
(trilema) 2017-03-11 asciilifeform: per dijkstra's 'tests reveal presence of bugs, but never their absence' |
00:44 |
verisimilitude |
Tests don't work, thehorrors. |
00:44 |
asciilifeform |
nor does 'coupled' refer to 'embedded in comments' |
00:47 |
verisimilitude |
Read this, thehorrors. |
00:47 |
thehorrors |
asciilifeform, I was thinking someting /like/ (conceptually) kelp, so not the comments exactly, but the separation of proof from source code. A separate linked entity. |
00:48 |
verisimilitude |
Consider how I write machine code: The program is entirely separate from its metadata, and both are entirely separate from the article extensively documenting them. |
00:49 |
verisimilitude |
This is the way to do it, certainly. |
00:51 |
thehorrors |
so what do you think about something like sqlite approach? do you think none of their testing works? I have huge respect for Dijkstra, but I don't think his remark means or proofs that "tests don't work" |
00:54 |
thehorrors |
iterative revelation of the presence of bugs may be an easier approach towards quality than proving the absence - is all I am trying to say |
| |
↖ |
00:57 |
verisimilitude |
I don't use sqlite directly, and don't care. |
00:57 |
verisimilitude |
The silly C language programmers love finding holes to plug. |
00:57 |
verisimilitude |
They never think to eliminate them entirely. |
00:58 |
verisimilitude |
The criticize Rust in their stupid justifications, and oddly never mention Ada. |
00:58 |
verisimilitude |
They criticize, that is. |
01:01 |
thehorrors |
I find the study of such projects useful nevertheless. |
01:02 |
verisimilitude |
http://logs.nosuchlabs.com/log/asciilifeform/2022-05-04#1099837 The only way to write good software is to have it be correct and able to be fast and then have it made to actually be fast. Most software isn't correct, but is able to be fast, and is then made to be slow in an infinitesimal stepping towards correctness. |
01:02 |
dulapbot |
Logged on 2022-05-04 20:52:27 thehorrors: iterative revelation of the presence of bugs may be an easier approach towards quality than proving the absence - is all I am trying to say |
01:02 |
verisimilitude |
I find masturbation useful, but I don't think others would necessarily want to watch. |
| |
↖ |
01:03 |
verisimilitude |
I understand how this looks, thehorrors. |
01:03 |
verisimilitude |
The sqlite authors are respected and experienced, whereas I'm not. |
01:03 |
verisimilitude |
I should be wrong then, right? |
01:03 |
verisimilitude |
Enough people can't be wrong for decades, right? |
01:06 |
thehorrors |
heh, you are putting words in my mouth. I am neither interested in promoting sqlite specifically nor trying to evaluate your proficiency. I said what I said - I find things interesting and worthy of examination, especially if they are not orthodox |
01:07 |
verisimilitude |
Surely I hit close to the mark, right? |
01:08 |
thehorrors |
I suppose I am still too new here not to be percieved as the guy who is trying to fuck with you and not simply someone who is trying to engage in productive discussion |
| |
↖ ↖ |
01:08 |
verisimilitude |
Don't feel that way; I fuck with everyone. |
01:10 |
verisimilitude |
Still, that surely hit close to the mark, I suppose. |
01:11 |
shinohai |
http://logs.nosuchlabs.com/log/asciilifeform/2022-05-04#1099846 <<< the former is just sad, the latter is spot on. |
01:11 |
dulapbot |
Logged on 2022-05-04 21:01:07 verisimilitude: I find masturbation useful, but I don't think others would necessarily want to watch. |
01:11 |
shinohai |
$vwap |
01:13 |
shinohai |
$vwap |
01:13 |
busybot |
The 24-Hour VWAP for BTC is $ 39137.50 USD |
01:14 |
thehorrors |
verisimilitude, which one exactly? My alleged hidden message that Hipp will eat you for breakfast? |
01:21 |
verisimilitude |
http://logs.nosuchlabs.com/log/asciilifeform/2022-05-04#1099853 |
01:21 |
dulapbot |
Logged on 2022-05-04 21:06:29 thehorrors: I suppose I am still too new here not to be percieved as the guy who is trying to fuck with you and not simply someone who is trying to engage in productive discussion |
01:24 |
thehorrors |
This made me think of the "intelligence-adjusted Dunbar's number": the number of people we are willing to recognize as "playing characters", not the number of people we are cognitively capable of recognizing as such. |
01:27 |
thehorrors |
or rather "compatibility-adjusted" maybe? I bet there is a distribution: people who are cognitively capable of fitting more than 200 people have inversely proportional predisposition to do so |
01:27 |
verisimilitude |
Sure. |
01:42 |
thehorrors |
However, if you delegate your memory of trust to systems like WoT - I bet it can scale actually. Exceptions apply, ofc |
01:45 |
thehorrors |
But does it need to scale? The more I think about that the less trivial this question becomes |
| |
~ 2 hours 16 minutes ~ |
04:01 |
signpost |
http://logs.nosuchlabs.com/log/asciilifeform/2022-05-04#1099853 << >> http://logs.nosuchlabs.com/log/asciilifeform/2022-03-20#1086367 |
04:01 |
dulapbot |
Logged on 2022-05-04 21:06:29 thehorrors: I suppose I am still too new here not to be percieved as the guy who is trying to fuck with you and not simply someone who is trying to engage in productive discussion |
04:01 |
dulapbot |
Logged on 2022-03-20 20:09:10 signpost: believes in the time-honored tradition of men giving each other shit as much as possible. |
04:04 |
signpost |
it's a delight to spar, and paying someone respect when you bother to. |
| |
↖ |
04:04 |
signpost |
literally, returning their gaze. |
04:09 |
signpost |
and not only, but telling them what you see, even if you don't like it. |
04:09 |
signpost |
*if they don't like it |
| |
~ 15 minutes ~ |
04:24 |
* |
signpost has meanwhile been putting together the init process for pentacle |
04:26 |
signpost |
I'll throw another WIP copy of it up on my blog tomorrow sometime for folks that want to test it in a chroot. |
04:29 |
* |
signpost built bitcoind with it the other day, having ported it and the deps (bdb, boost, openssl, etc) to pbuilds, worked nicely. |
04:30 |
signpost |
given the importance of bitcoind I see no reason to remove its own build process, but for the bitcoind pbuild I just make -f makefile.unix |
04:32 |
signpost |
pentacle can be thought of as a spiritual successor to rotor, tbh. I wanted sane and easily reproducible build environments for everything I use. |
| |
~ 15 minutes ~ |
04:47 |
thehorrors |
http://logs.bitdash.io/asciilifeform/2022-05-05#1099872 << Sure. There is also sometimes a fine line between sparring and trolling. I've never been interested in either much. There are plenty of opportunities to give each other shit while intelligently discussing ideas, in a much more interesting and subtle way. I suppose it's a matter of preference. |
04:47 |
bitbot |
Logged on 2022-05-05 04:04:00 signpost: it's a delight to spar, and paying someone respect when you bother to. |
04:50 |
thehorrors |
keyboard MMA on the other hand does not strike me as a particularly masculine, delightful or respectable activity. Maybe I am just a snob |
| |
~ 17 minutes ~ |
05:07 |
verisimilitude |
I'm not trolling. |
05:09 |
signpost |
thehorrors: I was telling you to hit him back. |
05:10 |
signpost |
instead of politely discussing the rules, say. |
05:11 |
signpost |
everyone's atomized into their own pocket of "interesting" enough to halt culture. |
| |
↖ |
05:13 |
signpost |
http://logs.nosuchlabs.com/log/asciilifeform/2022-03-03#1082377 << example of a recent thread where I didn't politely whine that the mods wouldn't remove someone I didn't find "interesting". |
05:13 |
dulapbot |
Logged on 2022-03-03 19:27:13 signpost: verisimilitude is signaling his status as a middle-caste slave with his nazi racial theory fantasies, lest he be lowered further. |
05:17 |
verisimilitude |
http://logs.nosuchlabs.com/log/asciilifeform/2022-05-05#1099887 This refers to those here, right? |
05:17 |
dulapbot |
Logged on 2022-05-05 01:09:31 signpost: everyone's atomized into their own pocket of "interesting" enough to halt culture. |
05:20 |
signpost |
as broadly as the word means. |
05:21 |
verisimilitude |
I agree, then. |
05:21 |
* |
signpost will grant that the amish probably still know how to talk to one another, but halted by other means. |
05:21 |
signpost |
collectively, I suppose. |
05:21 |
verisimilitude |
Each one of us is in his own world, but with enough tangents to stay. |
05:24 |
thehorrors |
verisimilitude, I never said you were, wasn't really saing anything about you specifically. It's just I was not sure what to tell you when I say something about sqlite unit testing and you respond with "oh you think I don't know how to use a computer? Huh? Yeah?" |
05:25 |
verisimilitude |
It's simple misprediction. |
05:25 |
verisimilitude |
Read what I write here in a monotone, and it will have the intended effect. |
05:38 |
signpost |
neither of you were communicating well. you quoted a bit from your hidden belief systems, said nothing substantial of the models informing the output. |
05:40 |
verisimilitude |
It would be a hard sell to tell thehorrors ``Read the entirety of my website.'' now wouldn't it? |
05:42 |
verisimilitude |
No, I'm not communicating well at all, right now. |
05:42 |
verisimilitude |
That was a non sequitur, wasn't it? |
05:54 |
thehorrors |
I am reading more than you might think perhaps. But I am taking it one step at a time. |
05:55 |
thehorrors |
I've read the entirety of loper-os and about half way through trilema |
05:56 |
verisimilitude |
That's a lot of reading. |
05:56 |
thehorrors |
Still trying to answer the question of "what exactly is the project?" I am getting conflicting messages from different sources |
| |
↖ |
05:56 |
thehorrors |
yes, I took a deep dive - just the way I like it |
05:57 |
thehorrors |
It unfolds to me unevenly, with a lot of discontinuities |
05:59 |
signpost |
this reminded me of a thread where phf and I had a mad. |
05:59 |
dulapbot |
(trilema) 2017-04-10 phf: you clearly have some twisted issues related to me. ~that was not my intent~. ~i genuinely think that you were speaking of pulling levers~ ~because there's nowhere in the thread where you attempted to dissuade me of my notion~ |
05:59 |
signpost |
complete failure to discuss the unstated parts of our disjunct models of the world. |
06:00 |
signpost |
kinda funny to read now. |
06:00 |
thehorrors |
signpost, taking the throughput of the channel into account - how can you hope for efficient model sync? |
06:02 |
signpost |
there's actually quite a lot of me in the logs at this point, so I imagine it's easier to assemble a me-simulation as time goes on, not harder. |
06:03 |
signpost |
of the assembled, I was probably the most tech-bro adjacent of the bunch, so phf's nose wasn't even wrong, if allergic. |
06:04 |
thehorrors |
It seems reductionist. Even though I read all those blog posts - I am nowhere near understanding who Stan or MP are. There was a poet who said something like "a though spoken is a lie" |
06:05 |
signpost |
I brought my "fuck you, I will discuss any topic I choose to discuss" and I can only imagine was his "casuals are mauling what was a high pursuit", and these had the expected result. |
06:06 |
signpost |
thehorrors: I'd expect the logs to be more illuminating there, and even that might be hard given the years of context. |
06:09 |
thehorrors |
well, yeah, that's another thing - the sheer amount of information that keeps accumulating over the years, with context. Good grief. I don't think anyone can read through it all. |
06:09 |
signpost |
http://logs.nosuchlabs.com/log/asciilifeform/2022-05-05#1099907 << a republic is a machine which evaluates this question for as long as it can. |
06:09 |
dulapbot |
Logged on 2022-05-05 01:54:54 thehorrors: Still trying to answer the question of "what exactly is the project?" I am getting conflicting messages from different sources |
06:09 |
thehorrors |
that contradicts MP however |
06:10 |
signpost |
say how? |
06:10 |
thehorrors |
he was all about "measurable outcome". Struggling to find that post in bookmarks |
06:11 |
thehorrors |
it was the post about utopias |
06:12 |
thehorrors |
http://trilema.com/2014/the-problem-of-ideal-social-systems-reprint |
06:12 |
signpost |
mp's notion that everyone not sucking on his toes was an NPC had a lot to do with limiting his impact on the world. |
06:12 |
thehorrors |
"concrete reference" he called it |
06:12 |
signpost |
but I don't see a contradiction in valuing measurable outcomes. |
06:13 |
thehorrors |
I tend to distill concrete ideas, I am not particularly fond of his (public) character |
06:15 |
signpost |
doesn't building one product inform the next one you build? |
06:15 |
signpost |
iterating on "what's the project" is what a business is. |
06:17 |
verisimilitude |
Has anyone seen a functional business, then? |
06:17 |
verisimilitude |
This seems more like the myth of a business. |
06:20 |
signpost |
depending on where you put the goalposts, I ran one for a decade. |
06:20 |
signpost |
might even fart out another, who knows |
06:21 |
thehorrors |
I got disconnected for a second. Anyways, the closest I've found is http://trilema.com/2012/gpg-contracts/ |
06:21 |
verisimilitude |
Fine, signpost. |
06:26 |
signpost |
thehorrors: http://logs.nosuchlabs.com/log/asciilifeform/2021-08-30#1054876 << this thread might have some useful context |
06:26 |
dulapbot |
Logged on 2021-08-30 20:49:40 pete_rizzo_: thanks for the invite |
06:26 |
thehorrors |
He had a business, other people in tmsr had businesses. I think tmsr was something else for him, something he abandoned in 2020; I've found his rationale unsatisfactory. |
06:26 |
thehorrors |
I'll read it |
06:26 |
signpost |
try to stop using hand-waving like "something else" |
06:27 |
signpost |
what else? |
06:27 |
signpost |
http://logs.nosuchlabs.com/log/asciilifeform/2021-08-31#1054946 << e.g. |
06:27 |
dulapbot |
Logged on 2021-08-31 06:46:41 punkman: even by end of 2014 some called it a "cult" http://www.contravex.com/2014/11/04/lets-cut-to-the-chase-is-la-serenissima-a-cult/ |
06:27 |
signpost |
I think the vision was as grand as reality would allow, and it didn't. |
06:28 |
thehorrors |
I'd tell you if I understood. Some sort of movement is my current understanding. |
06:29 |
thehorrors |
(clearly flirting with cryptoanarchy) |
06:32 |
signpost |
I'm not an anarchist, but a republican. |
06:32 |
thehorrors |
but also a sort of a mix between "bronze age", Plato, "might is right" and early Peterson-esque type of initiatives. |
06:33 |
thehorrors |
Oh I am not talking about you, signpost |
06:33 |
thehorrors |
I am talking about my attempt to understand what MP was about |
06:33 |
signpost |
afaik we preceded peterson's public life by years. |
06:34 |
signpost |
http://trilema.com/2018/the-republic-without-mp/#selection-182.0-182.1 << at some point mp turned from republicanism to this kind of monarchial thinking, continued to call it republicanism. |
06:34 |
signpost |
but also feudalism, because "eat shit, I will force you to embody contradiction" |
06:36 |
signpost |
that said I don't even dislike peterson. |
06:37 |
signpost |
he could stop crying so much while he's preaching traditionalism, but I've never quit benzos, so what do I know. |
06:41 |
thehorrors |
http://trilema.com/2018/the-republic-without-mp/#comment-124676 << hmm |
06:41 |
signpost |
http://logs.nosuchlabs.com/log-search?q=declaration&chan=asciilifeform << see also |
06:44 |
thehorrors |
yeah, I've seen that. I didn't realize it all went to feudalism at some point though. The declaration itself didn't not hint at that |
06:45 |
thehorrors |
so where did you come in, signpost? |
06:47 |
signpost |
http://logs.nosuchlabs.com/log/trilema/2014-10-16#879131 << here |
06:47 |
dulapbot |
(trilema) 2014-10-16 mircea_popescu: !up undata |
06:50 |
thehorrors |
eight years, huh |
06:51 |
thehorrors |
you seem to be critical of MP. And you implemented WoT, didn't you? |
06:52 |
signpost |
http://logs.nosuchlabs.com/log/trilema/2014-06-16#721072 << actually this preceded, apparently |
06:52 |
dulapbot |
(trilema) 2014-06-16 trinque: I can take it |
06:52 |
signpost |
thehorrors: I implemented maybe the 3rd or 4th iteration |
06:52 |
thehorrors |
who did the first one? |
06:53 |
signpost |
first one would've been the old bitcoin-otc WoT, I believe. |
06:53 |
signpost |
http://logs.nosuchlabs.com/log/asciilifeform/2021-08-31#1055007 |
06:53 |
dulapbot |
Logged on 2021-08-31 12:27:39 punkman: <punkman> at some point there was an incident with bitcoin-otc bot, and then the WoT split, and a new separate WoT was created, administered by kakobrekla/assbot |
06:54 |
signpost |
that whole punkman thread has a bunch of major events |
06:58 |
thehorrors |
But WoT is centralized, isn't it? |
06:59 |
signpost |
http://logs.nosuchlabs.com/log/asciilifeform/2021-07-18#1046518 << I agree that decentralizing it is important. |
06:59 |
dulapbot |
Logged on 2021-07-18 17:06:32 signpost: wants the decentralized wot, decentralized comms network, cryptographic control of what information leaks from my relationships |
07:00 |
thehorrors |
Is anyone building it? I want to build it |
07:01 |
signpost |
would be a mighty fine thing to build atop pest. |
07:01 |
signpost |
say as an extension of thimbronion's blatta |
07:02 |
thehorrors |
I didn't read the spec yet, but isn't pest a p2p communication protocol? How would it support decentralized WoT persistence? In my view it should be something like a blockchain |
07:02 |
thehorrors |
I don't understand blatta yet either - didn't look at it |
07:03 |
thehorrors |
is asciilifeform building pest? Is it in Ada? |
| |
↖ |
07:03 |
signpost |
it would support decentralized wot gossip. what the messages for wot updates look like, how they are stored, is not defined. |
07:04 |
signpost |
I'm skeptical you need an entire blockchain for this, since there's not much of a double-spend problem with statements about reputation. |
07:05 |
signpost |
(but it may entertain you to know that I wrote the wot for ethereum) |
07:05 |
signpost |
haven't released or anything. |
07:06 |
thehorrors |
it does entertain me, thank you |
07:06 |
signpost |
lol |
07:07 |
thehorrors |
the decentralized WoT will need more deliberation. It all depends on how people view it and want to use it. In my view a "global", visible wot would be best |
07:07 |
signpost |
I figure all pest-wot needs is a rule like "I believe the worst rating I receive with the most recent bitcoin block hash" |
07:07 |
thehorrors |
therefore, a "public ledger of trust" |
07:08 |
signpost |
and if I routinely had somebody in my wot emitting spam ratings with the same hash, I'd just negrate them. |
07:08 |
signpost |
eh, trust is not global. |
07:08 |
signpost |
that's entirely the wrong way to think about it, and leads directly to chinese social credit. |
07:09 |
thehorrors |
well then there is not much value in decentralizing it - everyone will just have their own little wot |
| |
↖ |
07:09 |
signpost |
one can only have trust along the specific paths across the social graph from me to you. |
07:09 |
signpost |
you're misunderstanding. global state yes, but what that state means *to you* is necessarily different than to me. |
07:09 |
thehorrors |
oh, sure |
07:10 |
signpost |
there's no such thing as "my score in the wot" |
07:10 |
thehorrors |
no, I agree |
07:10 |
signpost |
there's just "across these relationships I have these paths of transitive trust to thehorrors" |
07:10 |
thehorrors |
I mean more like "publicly visible" |
07:10 |
signpost |
yeah 100% agree with that |
07:10 |
signpost |
"res publica" |
07:11 |
thehorrors |
right, because I want to see if you trust that guy and that guy trusts the guy I trust etc |
07:11 |
signpost |
yup yup |
07:11 |
thehorrors |
ok |
07:11 |
thehorrors |
yeah, the scores are individual, just the way it is now |
07:11 |
thehorrors |
it's the decentralized storage I think is all that is needed pretty much |
07:14 |
signpost |
http://logs.nosuchlabs.com/log/trilema/2017-01-26#1607218 << worthwhile thread too, if I got the right one |
07:14 |
dulapbot |
(trilema) 2017-01-26 mircea_popescu: the fundamental problem here is that there's no way to reason in the manner you expect to reason. IF anything about the wot process is opposable to anyone involved ; then the wot process becomes by that measure less useful. no change whatsoever appears in the swamp it's made to confront. |
07:15 |
signpost |
mp raised the objection to signing ratings that the ratings become sworn statements opposable to the rater. |
07:16 |
signpost |
meaning legally, traditionally, and practically today before the megastate. |
07:18 |
signpost |
I think it's still worth discussing, because how the fuck do I trust that the relationship described by the rating actually exists without the signature in a decentralized setting with no elected trusted observer (e.g. deedbot, which verified the rater without sharing that verification with others) |
07:19 |
thehorrors |
I had a blasphemous idea about that |
07:20 |
verisimilitude |
http://logs.nosuchlabs.com/log/asciilifeform/2022-05-05#1099984 I'm, slowly. |
07:20 |
dulapbot |
Logged on 2022-05-05 03:01:52 thehorrors: is asciilifeform building pest? Is it in Ada? |
07:20 |
thehorrors |
Still not sure if it is any good. |
07:22 |
thehorrors |
you can assign public keys (and therefore trust) to systems, they don't have to be people. |
| |
↖ |
07:23 |
signpost |
how do you prevent your trusted system from becoming a dumping ground for people who aren't themselves held to account for what they did to it? |
07:24 |
thehorrors |
that's one of the questions, yes |
07:24 |
signpost |
imho this is why asciilifeform introduced V, to solve that problem |
07:24 |
thehorrors |
verisimilitude, I'll check your blog, you got patches? |
07:25 |
verisimilitude |
No, but I've source code. |
07:25 |
verisimilitude |
I'm building all components first. |
07:25 |
verisimilitude |
Nothing specific to Pest is yet present. |
07:26 |
thehorrors |
signpost, how do you see V solving that problem? |
07:26 |
thehorrors |
verisimilitude, what are you writing it in? |
07:27 |
verisimilitude |
I'll be writing it in Ada. |
07:27 |
signpost |
because there is no centralized authority on what authors I accept in my local V. |
| |
↖ |
07:28 |
thehorrors |
signpost, not sure I understand you. Suppose I want to rate you as X |
07:28 |
signpost |
I have a wot directory containing some public keys, and the tool assembles a source code state using that and the patches/sigs I have |
07:28 |
signpost |
I am addressing the idea of having systems hold a key, but perhaps you should explain what you hoped to solve with that idea. |
07:29 |
thehorrors |
I tell everyone "I am rating signpost as X" and I sign it. Everyone can verify that I signed it, so the relationship does exist, no? |
07:29 |
signpost |
yup |
07:29 |
thehorrors |
so what's the problem? |
07:29 |
thehorrors |
Noone can do this for me (fake my rating) |
07:29 |
signpost |
the problem is the gpg-contracts interpretation of signing |
07:30 |
signpost |
you're swearing to the veracity of the rating, such that it can hypothetically stand on its own legally, rather than act as a pointer which causes me to ask you personally if I should go do business with the person you rated. |
07:30 |
signpost |
the idea was to augment the human relationship, not replace it. |
07:31 |
signpost |
this might be a pointless religious notion standing in the way of something useful. I'm just saying what the past context was. |
07:32 |
signpost |
I've loosened up on my own interpretation of signing, having signed entire linux kernels/gccs/etc for pentacle |
07:32 |
signpost |
and holy shit, I never realized the lovecraftian terror of c standard headers before this project, not like this. |
07:33 |
thehorrors |
ok, well but then we can do without wot altogether to be honest. As long as everyone knows each other's keys, we can ask each other questions about the keys and be done with it. |
| |
↖ |
07:33 |
signpost |
asciilifeform once took a crack at enumerated meaning of signatures. |
| |
↖ |
07:33 |
signpost |
nah, I think computable trust is entirely valuable. think what it means at the level of packet routing, say. |
07:34 |
signpost |
suppose I lose my connection to asciilifeform, but I need to tell him something direly. it'd be nice if my machine had some sense of where to send that message, such that it will eventually reach him. |
07:34 |
thehorrors |
what is pentacle? is that a distro I am getting a sense? |
07:35 |
thehorrors |
For that you would have a server you trust, maybe? |
07:35 |
signpost |
http://www.loper-os.org/?p=1545 |
07:35 |
signpost |
yes, pentacle's a distro I'm working on; nope, it won't have anything but signed vpatches, and nobody's system will be identical, nor will it rely on somebody's "repo" of packages |
07:36 |
thehorrors |
how is it different from billymg's gentoo? |
07:37 |
signpost |
lifted a lot of inspiration from gentoo; I used it for many years, am on a gentoo machine right now. |
07:38 |
signpost |
where it differs is in beingg built to use V for src. |
07:39 |
thehorrors |
ah |
07:39 |
signpost |
*being |
07:41 |
signpost |
gentoo has also succumbed to significant complexity and interdependency creep. asciilifeform's frozen gentoo is great, but still requires a lot of work to cut back the dependency hell. |
07:41 |
signpost |
ends up being a littany of USE="-oh -no -you -fucking -dont" |
07:42 |
* |
signpost built pentacle outwards from what'd build the compiler itself, then "ok, now I need a kernel/init/text-editor" |
07:44 |
thehorrors |
have you looked into bsds? |
07:45 |
signpost |
sure, I ran openbsd for a long time |
07:45 |
signpost |
deedbot's server is openbsd |
07:45 |
signpost |
http://paste.deedbot.org/?id=qBYg << current pentacle kit |
07:45 |
verisimilitude |
I checked recently, and ``litany'' has but one letter t, signpost. |
07:46 |
signpost |
hm, ty |
07:46 |
verisimilitude |
It's no issue. |
07:46 |
verisimilitude |
I'd also thought it had two. |
07:47 |
thehorrors |
why not make a bsd distro instead of linux? |
07:48 |
signpost |
familiarity, probably. I don't think anybody'd object to someone making a set of BSD vpatches. |
07:49 |
thehorrors |
That's another project I will be thinking about. |
07:51 |
signpost |
I have a few state-management ideas I'm working on atop pentacle which involve squashfs and overlayfs, descendent of the way I used to do embedded linux for another project. |
07:51 |
signpost |
working on a formalized notion of a "job" which I can huck easily at a server running this system, and have it fire up and receive command/control via pest. |
07:52 |
signpost |
"job" here is basically a special signed-and-encrypted filesystem image. |
| |
↖ |
07:54 |
thehorrors |
well, here is where I disagree with heterogeneous systems. If you have identical systems, would you need such jobs? |
07:54 |
thehorrors |
it's a cool idea, don't get me wrong |
07:54 |
signpost |
you just buzz-worded at me, unpack the macros |
07:55 |
thehorrors |
well you said that no one's system will be identical, right? |
07:55 |
signpost |
my servers will certainly be identical. |
07:55 |
signpost |
yours will not be identical to mine, necessarily. |
07:56 |
thehorrors |
ok. Your concept of a job seems like a container, no? |
07:56 |
thehorrors |
why else would you need to exchange file systems instead of just saying "do this on this data"? |
07:57 |
signpost |
wtf does "do this on this data" mean aside "here's a binary, run and give it this input" ? |
07:58 |
thehorrors |
right |
07:58 |
signpost |
restate your question? |
07:58 |
thehorrors |
so why do you want jobs that entail exchanging file systems? |
07:59 |
signpost |
restate it again lol |
07:59 |
signpost |
at some point, you have a wad of program and data you want on some box |
07:59 |
signpost |
I am not spending the rest of my life polishing bespoke boxes by hand. |
07:59 |
thehorrors |
http://logs.bitdash.io/asciilifeform/2022-05-05#1100075 |
07:59 |
bitbot |
Logged on 2022-05-05 07:52:55 signpost: "job" here is basically a special signed-and-encrypted filesystem image. |
07:59 |
signpost |
the things are getting a kernel installed on them and becoming part of a compute fabric I own. |
08:00 |
signpost |
and I'm not using fucking kubernetes to do that. |
08:00 |
thehorrors |
all I am trying to figure out is why do you want such "jobs" as you define them |
08:01 |
signpost |
explain what "heterogeneous systems" are to you, and why you disagree with them |
08:02 |
thehorrors |
"I have a different version of everything", practically meaning "you cannot assume what version of binaries etc I have (or a server has)" |
08:02 |
signpost |
you're misunderstanding. |
08:04 |
signpost |
when I say pentacle uses V as a fundamental principle, I mean that it's a machine into which *you* can load different vpatches than *me*. |
08:04 |
thehorrors |
right |
08:04 |
signpost |
thus we might mean very different things when we say we both run "pentacle" |
08:06 |
signpost |
when I'm discussing what I want to do with my servers, that assumes a uniform node substrate I build (and probably call a distinct piece of software from pentacle) which allows me to declaratively load jobs on any of the machines in my fleet. |
08:06 |
thehorrors |
okay. And all the servers you control will have the same vpatches. |
08:06 |
thehorrors |
yep |
08:07 |
thehorrors |
the only detail here I want to clarify is the jobs themselves |
08:07 |
signpost |
the bitcoind job might have a statically linked bitcoind in it, and a script that fires it up. |
08:07 |
signpost |
maybe a ML job has a python and pytorch in it. |
08:08 |
thehorrors |
so my question was "why do you want to package jobs as filesystems" |
08:08 |
signpost |
programs are files in *nix |
08:08 |
signpost |
what's the alternative? |
08:09 |
thehorrors |
send a server a binary and data, an arhive, basically |
08:09 |
thehorrors |
why bother with a filesystem? |
08:10 |
signpost |
you're just giving an alternative without the merits of the alternative, but I'm happy to explain why I like this approach better. |
08:10 |
signpost |
but I'm running out of steam, so after this gonna hit the hay |
08:11 |
thehorrors |
that's all I was asking for. That's okay, we both might be getting a bit tired. GN, signpost, good talking to you |
08:11 |
signpost |
an archive is just a shitty filesystem image which you have to unpack somewhere, take up even more disk space with that. |
08:11 |
signpost |
then when you read it, it'll end up in the linux fs cache anyway in RAM |
08:12 |
signpost |
seems like mounting a squashfs is less duplicative. |
08:12 |
thehorrors |
I see, it has merit |
08:13 |
signpost |
then the use of overlayfs is for composition. two jobs might both want busybox's utils around for shell. |
08:15 |
signpost |
can have one busybox.squashfs, and union-mount w/e's different atop. |
08:19 |
thehorrors |
that explains it, thank you |
08:21 |
* |
signpost quite open to better ways, so long as same-or-better management of the linux system-state shitwad. |
08:22 |
signpost |
alrighty, gnite. |
08:22 |
thehorrors |
gnit, signpost, thanks! |
| |
~ 1 hours 4 minutes ~ |
09:27 |
crtdaydreams |
http://logs.nosuchlabs.com/log/asciilifeform/2022-05-05#1100032 << >> http://logs.nosuchlabs.com/log/asciilifeform/2022-01-21#1075607 |
09:27 |
dulapbot |
Logged on 2022-05-05 03:26:06 signpost: because there is no centralized authority on what authors I accept in my local V. |
09:27 |
dulapbot |
Logged on 2022-01-21 20:44:32 asciilifeform: if a hero arises who comes up with an algo for forcing handles on a multi-level pestnet to actually stay unique -- let's all take off hats & use it |
09:30 |
crtdaydreams |
I've been thinking about this. What signpost mentions is a very similar problem regarding a decentralized WoT without an agent to say "this key belongs to xyz" and verify that. |
09:31 |
crtdaydreams |
2 Solutions, one more novel than the other. |
09:34 |
crtdaydreams |
Firstly, it's important to make a distinction between a username and nick. Think in terms of IRC usernames e.g. cdd@192.168.1.1, but my nick is crtdaydreams |
09:34 |
crtdaydreams |
username is a network identifier, a nick is just a handle |
09:46 |
crtdaydreams |
1) Arguably less sophisticated solution, but werks. All peers are identified by their fingerprint and the entire notion of handles is purely client-side. When peering the peer will submit a handle, if the handle is already taken within your network the peers handle will default to "handle_fingerprint", from the client side you can then change the client-side handle of that |
| |
↖ |
09:46 |
crtdaydreams |
peer. If the /nick command is used by a peer in a chan, the same principle applies. |
09:52 |
crtdaydreams |
2) Employs the use of a wot in which several peers can sign to say that "x key belongs to y handle" which can be automated. A simple "majority rules" would determine whether or not that peer can lay claim to that nick. |
| |
↖ ↖ |
09:53 |
crtdaydreams |
^ would revert to above to resolve a name conflict wherein all partys hold the same trust for that nick |
09:56 |
crtdaydreams |
Now that I mention it, it feels too trivial, like I'm missing something. If tis' been thought of and rejected would like a link in logz. |
| |
~ 3 hours 57 minutes ~ |
13:54 |
asciilifeform |
http://logs.nosuchlabs.com/log/asciilifeform/2022-05-05#1100133 << think about it for a minute. it doesn't matter what kind of string is used 'in place of handle', a l2 or more distant station can still collide (deliberately or otherwise) and you have precisely same problem as when using handles strictly |
13:54 |
dulapbot |
Logged on 2022-05-05 05:45:14 crtdaydreams: 1) Arguably less sophisticated solution, but werks. All peers are identified by their fingerprint and the entire notion of handles is purely client-side. When peering the peer will submit a handle, if the handle is already taken within your network the peers handle will default to "handle_fingerprint", from the client side you can then change the client-side handle of that |
13:56 |
asciilifeform |
http://logs.nosuchlabs.com/log/asciilifeform/2022-05-05#1100135 << there is no public key crypto in pest. pest keys are secret keys (i.e. known strictly to a given pair of peers), your statement makes 0 sense in context of pest |
13:56 |
dulapbot |
Logged on 2022-05-05 05:51:21 crtdaydreams: 2) Employs the use of a wot in which several peers can sign to say that "x key belongs to y handle" which can be automated. A simple "majority rules" would determine whether or not that peer can lay claim to that nick. |
13:57 |
asciilifeform |
crtdaydreams: plz read the spec. |
13:58 |
asciilifeform |
crtdaydreams: ... and the various threads re why no rsa in pest. |
13:58 |
dulapbot |
Logged on 2022-01-24 11:58:52 asciilifeform: thimbronion: recall the orig. reason why pest is using symmetric crypto, btw |
14:02 |
asciilifeform |
http://logs.nosuchlabs.com/log/asciilifeform/2022-05-05#1099997 << that's what even ~means~ to decentralize -- yes, 'everyone will have own little wot', which is exactly as it should be. there is no actual reason why wot needs 'universally agreed global state' at all. |
14:02 |
dulapbot |
Logged on 2022-05-05 03:07:30 thehorrors: well then there is not much value in decentralizing it - everyone will just have their own little wot |
14:03 |
asciilifeform |
observe that nuffin stops you from propagating the particular ratings you can see from your l1 to l2+ on request. |
14:03 |
dulapbot |
Logged on 2022-03-25 20:03:40 asciilifeform: (nuffin stops a peer from later adding the warez to his own share and then pass to ~his~ l1. is the logical way to propagate it.) |
14:05 |
asciilifeform |
http://logs.nosuchlabs.com/log/asciilifeform/2022-05-05#1100021 << and what happens when someone walks off with (or via other means lifts the key from) the box ? or finds a way to get it to sign $arbitrary turd. |
14:05 |
dulapbot |
Logged on 2022-05-05 03:20:45 thehorrors: you can assign public keys (and therefore trust) to systems, they don't have to be people. |
14:06 |
asciilifeform |
mechanically-firing pubkey signers are a braindamaged concept. |
14:06 |
asciilifeform |
( and walks straight into 'dao'-style lulz ) |
14:12 |
asciilifeform |
http://logs.nosuchlabs.com/log/asciilifeform/2022-05-05#1100046 << a decentralized wot is precisely a means for 'asking questions about keys' in such a way that one can post, for one's l1, the answers at a given time. |
14:12 |
dulapbot |
Logged on 2022-05-05 03:31:26 thehorrors: ok, well but then we can do without wot altogether to be honest. As long as everyone knows each other's keys, we can ask each other questions about the keys and be done with it. |
14:15 |
asciilifeform |
this is the correct way to implement a wot, such that it doesn't reduce to 'some d00d's www' which lives & dies with him; or to the ancient meatspace snail grind where you gotta write to yer l1, and ask'em to write to ~theirs~, and wait months/years for a complete picture |
14:16 |
asciilifeform |
it requires, yes, a usable secure p2p comms protocol, to implement. which at last exists. |
14:17 |
asciilifeform |
( and in which, observe, the 'root of trust' is still manually-operated rsa signatures . as it oughta be. ) |
| |
~ 1 hours 22 minutes ~ |
15:39 |
asciilifeform |
crtdaydreams: see also e.g. this thrd re the essence of the problem of unique handles |
15:39 |
dulapbot |
Logged on 2022-01-19 22:02:03 asciilifeform: chains to begin with were asciilifeform's notion of how to give ~some~ measure of authenticity for l2+ hearsays |
15:41 |
asciilifeform |
asciilifeform's current notion for a pill ftr. |
15:41 |
dulapbot |
Logged on 2022-01-26 18:48:29 asciilifeform: thimbronion: current notion, summarized: we bite off 64byte from payload, and get 2 new fields, call'em 'unlock' and 'lock'. lock == h256(errything else in msg, incl. 'unlock', and unrevealed 32byte turd 'S'.); unlock = 'S' from yer previous msg. |
15:43 |
asciilifeform |
^ ugly but asciilifeform suspects it is closest thing physically possible to a solution |
| |
~ 1 hours 1 minutes ~ |
16:44 |
verisimilitude |
I suppose that would make the linked-list a doubly-linked-list, yes. |
16:50 |
asciilifeform |
verisimilitude: not precisely (in principle someone can forge a forward linkage once he has an old 'S'; however the forged chain is then ~distinguishable~, which is the point |
16:50 |
dulapbot |
Logged on 2022-01-26 12:23:03 asciilifeform: and of course dr.evil can still forge msgs in transit, so long as 1) he remembers, for all time, to withhold all further upstream ones 2) he is willing to be that his upstream victim has no other paths to the downstream dupes |
16:50 |
asciilifeform |
) |
16:54 |
verisimilitude |
Yes, but forgery is already trivial, right? |
16:55 |
asciilifeform |
verisimilitude: aha |
16:55 |
asciilifeform |
aim there is to make all l2+ speakers ~distinguishable~ |
| |
↖ |
16:55 |
verisimilitude |
If a packet drops correctly, someone can already impersonate others. |
16:55 |
asciilifeform |
verisimilitude: nobody can impersonate an l1 station (unless its key compromised) |
16:55 |
asciilifeform |
the issue is w/ l2+ |
16:56 |
verisimilitude |
The L1 refers to those stations with registered keys, right? |
16:56 |
asciilifeform |
verisimilitude: refers to a given station's direct peers |
16:56 |
verisimilitude |
So yes. |
16:56 |
verisimilitude |
Okay; a packet gets dropped, but I get the copies; impersonation can happen here. |
16:57 |
asciilifeform |
aint clear what dropped packets have to do with it. atm anyone can emit anyffin at all marked as 'rebroadcast' |
16:58 |
verisimilitude |
I'll write more about Pest when I'll be closer to finishing my client. |
16:58 |
asciilifeform |
verisimilitude: didja read the spec ? |
16:59 |
verisimilitude |
Yes. |
16:59 |
asciilifeform |
aite |
16:59 |
asciilifeform |
wb phf |
17:03 |
phf |
was going to opine on wot, but actually gtg |
17:03 |
asciilifeform |
a ok |
17:04 |
verisimilitude |
wtf wgl lol brb |
17:07 |
asciilifeform |
eh it aint as if there's some kinda hurry. let fella fix his tractor, feed horse, etc lol |
| |
↖ ↖ |
17:07 |
verisimilitude |
idk wym ymmv |
17:20 |
* |
signpost holds power button on verisimilitude down for 5sec. |
17:21 |
asciilifeform |
lol |
17:21 |
* |
asciilifeform installed an old-fashioned 'knife switch' mains switch yest., makes very satisfying 'thunk' |
17:24 |
verisimilitude |
Abbreviations being used so much is bothersome. |
17:26 |
verisimilitude |
Anyway, I repeatedly fail my self-imposed deadlines; my new goal for finishing a tiny little language service is within this month. |
17:28 |
verisimilitude |
I don't even like toki pona, but I'm nowhere near being able to write a Whitaker's Words program for Latin yet. |
17:30 |
verisimilitude |
I don't know how to write a CGI program, but there's a solution to that: I'm just going to expand the table to its passive form and add some URL rewriting rules to the Apache configuration. Even for Latin, it shouldn't be larger than one hundred mebibytes. |
| |
~ 23 minutes ~ |
17:54 |
asciilifeform |
ugh |
18:02 |
shinohai |
jfc killfiles can't happen soon enough. |
18:02 |
shinohai |
$vwap |
18:02 |
busybot |
The 24-Hour VWAP for BTC is $ 39137.50 USD |
18:10 |
verisimilitude |
I'm that bad, am I? |
| |
~ 2 hours 22 minutes ~ |
20:33 |
mats |
yes |
20:34 |
asciilifeform |
lol |
20:34 |
asciilifeform |
wb mats |
| |
~ 47 minutes ~ |
21:21 |
verisimilitude |
I just wanted to nudge the topic back to programming. |
21:22 |
verisimilitude |
Does anyone care to discuss tries, then, or something else? |
| |
~ 1 hours 5 minutes ~ |
22:28 |
phf |
http://logs.nosuchlabs.com/log/asciilifeform/2022-05-05#1100186 << it was the second, feed horse |
22:28 |
dulapbot |
Logged on 2022-05-05 13:05:32 asciilifeform: eh it aint as if there's some kinda hurry. let fella fix his tractor, feed horse, etc lol |
22:40 |
signpost |
dang, my mother just died of a heart attack. |
| |
↖ |
22:41 |
signpost |
pretty surprising, not much led up to it. |
22:46 |
phf |
signpost: please accept my condolences |
22:47 |
signpost |
thank you phf. |
22:51 |
shinohai |
Damn signpost my condolences as well. :/ |
22:53 |
signpost |
shinohai: thanks bud |
22:58 |
thimbronion |
I am so sorry signpost. |
| |
~ 23 minutes ~ |
23:21 |
billymg |
very sorry to hear that signpost, my condolences |
23:25 |
verisimilitude |
That's unfortunate to read, signpost. |
23:37 |
scoopbot |
New article on A Syndication of Verisimilitudes: A Trivial Trie in Ada |