?

Log in

No account? Create an account
A radical thought - Baxil [bakh-HEEL'], n. My Sites [Tomorrowlands] [The TTU Wiki] [Photos]
View My LJ [By Tag]


July 21st, 2008
05:28 pm
[User Picture]

[Link]

Previous Entry Share Next Entry
A radical thought
The hippies used to say "Don't trust anyone over 30." But, being hippies and not computer programmers, they left a surprising amount of syntactical ambiguity in that statement. Even if we don't challenge the assumption that "thirty" refers to the number of Earth solar years since the target's birth, this still leaves the question of what the exact cutoff is, and to hold any sort of rational discussion about the topic, it's necessary to find a rigorous answer.

Linguistically, "over" implies strictly "greater than"; if "greater than or equal to" were intended, a phrase such as "at least" would have been more appropriate. So if we were writing pseudocode for a trust routine, it should look like:

if ($person.age_in_years > 30) then { $person.trust = 0; }

Even this doesn't answer the question: What is the precision of the comparison? Is someone "over" thirty when they are 30 plus a month? 30 plus a day? 30 plus an hour, a minute, a second, a nanosecond?

Tests of arbitrary precision could easily be constructed and executed, but this would then create a secondary dilemma: platform differences would not guarantee consistent trust results across runtime environments. For full RFC compliance, and to prevent security exploits in applications implementing trust webs, we must ensure that all interpersonal relationships are calculable in replicable ways regardless of native hardware. Removing the question of precision is recognized as the canonical response to this issue:

if ( int($person.age_in_years) > 30) then { $person.trust = 0; }

Which is a lengthy way of saying that, as of yesterday, you can no longer trust me.

--
Edited to add: No, I didn't just turn 30. ];=8) If int(baxil.age_in_years) needs to be strictly greater than 30, then 30 > 30 fails. Besides, I already turned 30 last year.

Current Location: ~spiral
Current Mood: geekyolder
Current Music: "Spy", Cowboy Bebop OST 'Vitaminless'
Tags:

(25 comments | Leave a comment)

Comments
 
(Deleted comment)
[User Picture]
From:baxil
Date:July 22nd, 2008 07:12 pm (UTC)
(Link)
Yeah, I remember reading that in Cory Doctorow's "Little Brother." Have you spotted the slogan anywhere else?
[User Picture]
From:lysana
Date:July 22nd, 2008 07:38 pm (UTC)
(Link)
I've seen a Prius with the bumper sticker, "Don't trust anyone under 30 MPG."
[User Picture]
From:baxil
Date:July 23rd, 2008 01:03 am (UTC)
(Link)
Nice!
(Deleted comment)
[User Picture]
From:zuki_san
Date:July 22nd, 2008 01:42 am (UTC)
(Link)
Congratulations!
[User Picture]
From:todd_dharken
Date:July 22nd, 2008 01:49 am (UTC)
(Link)
I prefer the old X-Files slogan, "Trust no one."

It's even politically correct as it's not age discriminatory.

By the way, happy birthday to you.
[User Picture]
From:soreth
Date:July 22nd, 2008 01:59 am (UTC)
(Link)
Sure, but... when did I ever trust you? ;)
[User Picture]
From:dogemperor
Date:July 22nd, 2008 04:08 am (UTC)
(Link)
Congrabulations on getting another year closer to the Granny Cart, Draconic Edition. :D (Just hit my "do not trust + 5" a few days back, so it's all good there. :3)
[User Picture]
From:krinndnz
Date:July 22nd, 2008 05:13 am (UTC)
(Link)
Happy birthday, excellent dracoform.

As for hitting thirty, think of it this way - in JRPG years, you're a Revered Elder!
From:silussa
Date:July 22nd, 2008 06:45 am (UTC)
(Link)
Sounds more like you want condolences then hatching day greetings. :)

And for what it's worth, I passed 30 a LONG time ago.
From:drake [begriffli.ch]
Date:July 22nd, 2008 08:33 am (UTC)
(Link)

Are you sure you're not mixing up hippies talking about people with cypherpunks talking about asymmetric cryptographic keys? ;-)

(Or symmetric ones, for that matter.)

From:drake [begriffli.ch]
Date:July 24th, 2008 11:59 am (UTC)
(Link)

Oh, and, theoretically 30 years plus the Planck time. You could probably make “years” mean “years TAI”; you could remove leap years from the calendar to make it more absolute. Whether that is acceptable in this circumstance seems contentious, but it would certainly be more convenient to define it as (365 × 86400 × 9192631770 periods of the radiation for the hyperfine transition in the ground state of a cesium-133 atom at rest at mean sea level on Earth), but then if you start trying to take relativistic effects into account it gets very difficult anyway.

That, of course, is completely ignoring not only the limited precision of human timekeeping and the aforementioned relativistic annoyances, but the problems with “time of birth” for a human-bodied creature not being well-defined to that precision, as far as I know (nor even to nanosecond precision). One might try to define in some exact enough way what atoms constitute the contiguous mass of the newborn, define (by some cell structure predicate?) what cells of the biological mother constitute the boundary of the vagina to the outside world, approximate the threshold as a plane, then define the first intersection of the two as time of birth, but that doesn't completely work in the face of anything but trivial cases, even discounting totally different cases like Caesarean sections, each of which might have to be handled on its own. This is also ignoring the inconvenient contiguity of bodily function until the detachment of the umbilical cord.

So realistically I think the precision of measuring time of birth will be constrained to a resolution of about one second, assuming all the timepieces are calibrated well and everything proceeds quickly; if you can't assume the presence of a reliable observer with a reliable timepiece to that extent, or if the situation is more complicated, the resolution may be more on the order of ten seconds or a hundred seconds, or even arbitrarily coarser depending on just how unreliable the observation situation is.

So that actually yields a more probabilistic approach. (This is getting to where my knowledge gets flaky, so maybe this is totally wrong.) You take a measurement of the supposed time of birth to as much precision as you have available, then apply all the inaccuracies and compute a normal curve or somesuch that's the distribution corresponding to the available knowledge of time of birth; then you shift that forward 30 TAI years plus one Planck time. (If you need leap years, calculate them based on the original measurement, but if you insist on leap years then the case of humans born on February 29 is left as an exercise for the reader.) Define a threshold of confidence for how known it must be that at least 30 TAI years have passed since birth, then forward to that threshold in the cumulative distribution function and that becomes the greater-than-or-equal threshold for an infinitely precise timepiece. Then the current confidence given a current imprecise timepiece will be based on the, uh, hmm. *mumble, mutter, draw diagrams* Uh, the… integral over time t from −∞ to +∞ of timepiece-pdf(t) × threshold-cdf(t), divided by the same integral of just timepiece-pdf(t)? I think? Wait, you can remove the division—the integral of a probability distribution is required to be 1, isn't it? Guh.

And of course that's still semi-assuming relativity doesn't totally scuttle your measurements by creating dilation effects beyond some function of the accuracy of your timepieces. I think.

*head explodes*

[User Picture]
From:supersniffles
Date:July 22nd, 2008 10:12 am (UTC)
(Link)
Happy Birthday!!!!
[User Picture]
From:dirac
Date:July 22nd, 2008 11:02 am (UTC)
(Link)
Happy Birthday, Bax!

You're only 30? The way you present yourself, I always assumed that you were older physically, but then, it's because you're a wise old spirit, aren't you? ^^
[User Picture]
From:baxil
Date:July 22nd, 2008 07:32 pm (UTC)
(Link)
Heh. I heard that ALL the time when I was in my teens and collegiates. Nice to hear I can still surprise people with my outer age despite the bald spot and grey hairs. ];=8b

(Edited for usericon matching.)

Edited at 2008-07-23 01:04 am (UTC)
[User Picture]
From:wy
Date:July 22nd, 2008 12:36 pm (UTC)
(Link)
40 is the new 30. Nowadays, it's "don't trust the damned baby boomers"
[User Picture]
From:draque
Date:July 22nd, 2008 02:31 pm (UTC)
(Link)
You're certainly lucky that client side encryption will help you to protect data from yourself when you're transferring thoughts from one mental thread to another. Also, setting a boolean with a numeric value is bad... If you weren't so old, I would try to explain why.
[User Picture]
From:baxil
Date:August 5th, 2008 09:34 pm (UTC)
(Link)
You know, clearly I am getting old here, because it took me two weeks to re-read this and say, "Waaaait a second ..."

Trust isn't a boolean value! At least, in any sufficiently robust application, I should hope not. I might trust someone to, say, tell me their real name when posting anonymously, but not to vouch for someone's public key.

Of course, trust is probably best expressed as a multidimensional array (do I trust them to cook dinner for me? to correct logic errors in a mathematical proof? to take $300 in cash to the bank?), so the pseudocode probably should have accounted for that and just set trust to some Zero() function ...
[User Picture]
From:draque
Date:August 5th, 2008 09:49 pm (UTC)
(Link)
it's probably easiest to just store in a hash map of the trust subjects tied to an enum, if you really want to do it cleanly...
[User Picture]
From:joysweeper
Date:July 22nd, 2008 03:20 pm (UTC)
(Link)
*blinks* You are old, Baxil. *is ridiculously young*

Happy birthday!
[User Picture]
From:yaochi
Date:July 22nd, 2008 03:46 pm (UTC)
(Link)
What the hippies said was too true.

They are all well over 30 now and run the state of California.

Right into the ground it seems.

Trust no one over 30 or 25 ?

As stated above, "Trust No One".
[User Picture]
From:r_caton
Date:July 22nd, 2008 04:20 pm (UTC)
(Link)
The Truth Is Up There.

Some would never trust dragons and their riddling talk anyways....

Happy Hatchies! I'm past my half century.....
[User Picture]
From:mr_silvers
Date:July 22nd, 2008 04:37 pm (UTC)
(Link)
Congratulations on achieving a milestone that many do not.

And may many more come you way with the cheerfulness that accompanies them.
[User Picture]
From:lysana
Date:July 22nd, 2008 07:39 pm (UTC)
(Link)
Happy birthday!
Tomorrowlands Powered by LiveJournal.com