2023-03-03 06:14:39 <veltas      > The problem I'm grappling with today is backup. I have like 1TB of photos I want to safely backup
2023-03-03 06:15:05 <veltas      > I looked up offset backup solutions from tarsnap to Google One, everything is £15/month or more for that quantity
2023-03-03 06:15:42 <veltas      > So I have realised the most cost effective method is to just buy an external hard drive and just leave it at someone else's house, occasionally checking drive health and file integrity
2023-03-03 06:16:20 <veltas      > That's my offsite backup solution, it will cost like £50 or less one-off, and then have tiny travel costs and a little time investement in future
2023-03-03 06:17:08 <veltas      > Every online paid service is horrifyingly overpriced, even consumer-grade stuff.
2023-03-03 06:17:28 <veltas      > "offset backup" *offsite backup
2023-03-03 06:17:35 <veltas      > offset backup doesn't sound very useful lol
2023-03-03 07:35:00 <veltas      > Actually cheapest I've found is MEGA, for like 1.56 euros a TB a month
2023-03-03 07:35:39 <veltas      > I was just thinking how cheap can I go, if I started a business to do this myself, I think that's below what I can do at my scale
2023-03-03 07:37:45 <veltas      > It's still more expensive than buying a drive and doing it myself, but much more convenient and a lot less work
2023-03-03 07:45:44 <KipIngram   > The only problem I see with that is that you'd be trusting a single vendor; trusting their competence and their business longevity.
2023-03-03 08:05:35 <veltas      > KipIngram: It's not an issue though because it's a backup. I just have to trust them enough that they're available when I lose data on my end
2023-03-03 08:05:51 <veltas      > But to be fair, to do this right I need to check my side as well
2023-03-03 08:06:17 <veltas      > Which is already enough work that I might as well just self-host
2023-03-03 08:06:49 <veltas      > I think I want to go away and try and calculate how much this stuff actually costs
2023-03-03 08:06:58 <veltas      > if I wanted to make a company out of it
2023-03-03 08:11:08 <veltas      > There's no such thing as a serverless website
2023-03-03 08:44:23 <KipIngram   > Yeah, I do understand.
2023-03-03 08:44:48 <KipIngram   > And it's also not a "life or death" thing - it would suck to lose the pictures, but it wouldn't fundamentally ruin you.
2023-03-03 09:12:44 <KipIngram   > You could of course approach something like this by setting up your own facilities and buying hardware, but another possibility (once you had some volume) would be to acquire the storage from the cloud but at a cheaper cost via volume discounts; your profit would be in that volume discount zone.  That would spare you all the business of managing hardware yourself, having the risk of hardware failures,
2023-03-03 09:12:46 <KipIngram   > vandalism/theft, and so on.
2023-03-03 09:13:22 <KipIngram   > But that seems like a hard model to "get launched," because you'd need the volume to make it cost effective.
2023-03-03 09:28:25 <KipIngram   > Man, those damn silmirils sure caused a heap of trouble for the elves...
2023-03-03 09:50:38 <KipIngram   > By the way, guys - "geometric algebra" is a fascinating bit of math, and is something that isn't generally taught by the standard education program.
2023-03-03 09:51:04 <KipIngram   > Turns out it's a general approach that captures a whole bunch of things that we ARE often taught and "unifies" them.
2023-03-03 09:51:32 <KipIngram   > Complex arithmetic, vector/matrix stuff, quaternion theory, etc. - all are just subsets of the general machinery of geometric algebra.
2023-03-03 09:51:48 <KipIngram   > I really think it's the way we should be taught this stuff, because of this "universal coverage."
2023-03-03 09:53:08 <KipIngram   > Instead, though, we're taught complex arithmetic by *defining* i as the square root of -1, and working from there.  We're taught vector/matrix algebra as an isolated thing, and the cross-product is introduced ad hoc to cover the one bit of stuff you need from the general approach.  And we're usually not taught quaternions at all.
2023-03-03 09:53:23 <KipIngram   > But all of it just "falls out" of geometric algebra, in a very beautiful way.
2023-03-03 09:54:15 <KipIngram   > The only reason we can get away with the cross product trick is because we happen to be working in 3D; in higher dimensionality you can't represent "that thing" using a vector.
2023-03-03 09:55:15 <KipIngram   > what it's really capturing is a product of two vectors, which in general is a "bivector" - an order 2 thing.  But because 2 happens to be 3-1 (full dimensionality minus dimensionality of vector), you can use a specially chosen vector to stand in for the bivector.
2023-03-03 09:55:22 <KipIngram   > And that's where the cross product comes from.
2023-03-03 09:55:39 <KipIngram   > It's the bivector's "dual."
2023-03-03 10:00:30 <KipIngram   > In N dimensions you have scalars, vectors, bivectors, ... (N-1)vectors.  And you can add and multiply anything with anything.
2023-03-03 10:01:04 <KipIngram   > Generally the orders just add, modulo N.  So an (N-1)vector multiplied by a vector wraps you back around to a scalar.
2023-03-03 10:01:39 <KipIngram   > And the dual of an (M)vector is always an (N-M)vector.
2023-03-03 10:05:56 <KipIngram   > If you wanna know more about it, this is pretty good:
2023-03-03 10:05:58 <KipIngram   > https://www.youtube.com/playlist?list=PLpzmRsG7u_gqaTo_vEseQ7U8KFvtiJY4K
2023-03-03 10:06:52 <thrig       > as long as they keep the discussion clear and euclid
2023-03-03 10:20:49 <JITn        > When was that ""geometric algebra"" written?  Yesterday?
2023-03-03 10:23:04 <JITn        > I saw some YT presentation.  I am not impressed.  I guess it is beyond me.
2023-03-03 11:33:11 <KipIngram   > This is the paper in which Feynman first put forth is "path integral" formulation of quantum mechanics:
2023-03-03 11:33:13 <KipIngram   > https://puredhamma.net/wp-content/uploads/Feynman-R.P.-Space-Time-Approach-to-Non-Relativistic-Quantum-Mechanics-1948.pdf
2023-03-03 12:10:26 <KipIngram   > Wow - Feynman really was brilliant.
2023-03-03 12:11:20 <KipIngram   > JITn: Well, what I found excellent about it is how "straightforwardly" it sweeps up a whole array of topics that are normally introduced as independent branches of things.
2023-03-03 12:11:37 <KipIngram   > To my way of thinking, that kind of unificaiton is always good.
2023-03-03 12:12:05 <thrig       > rebooting instructional design might need some study before being done
2023-03-03 12:12:34 <KipIngram   > I assume that the educational system did things the way they did because a) only a subset of the total  knowledge was regarded as "necessary to convey," and b) they felt that independent presentations of those subset items was "easier" than learning the general machinery.
2023-03-03 12:12:45 <KipIngram   > thrig: I'm sure.
2023-03-03 12:12:58 <KipIngram   > Never good to change things in a knee-jerk way.
2023-03-03 12:13:44 <KipIngram   > I'll say it this way - if one's goal was to learn more or less the whole scope of stuff that the approach eventually accesses, then I definitely think using it as the instructional pathway would be preferable.
2023-03-03 12:14:20 <thrig       > but throw universals at kids and they'll probably be all like wut
2023-03-03 12:14:20 <KipIngram   > But, if your goal is indeed a limited subset of that total knowledge base, then arguing that an approach that avoids having to learn the extra machinery is not necessarily unreasonable.
2023-03-03 12:14:56 <KipIngram   > Well, the universal machinery isn't THAT hard; I think an average class could handle it.  But, I concede it's reasonable to debate it.
2023-03-03 12:15:10 <KipIngram   > Maybe I feel that way because it makes sense to *me*.
2023-03-03 12:15:21 <KipIngram   > I.e., now that I've worked through the machinery and "got it."
2023-03-03 12:15:49 <KipIngram   > And I only know how it feels to be a learner "like me" - I've only walked that path in one set of shoes.
2023-03-03 12:17:07 <KipIngram   > Really, though, the whole business relies on nothing beyond basic algebra.
2023-03-03 12:17:42 <KipIngram   > So it feels to me that anyone who'd done a decent job in first year algebra could "handle it."
2023-03-03 12:18:22 <KipIngram   > I'm distinguishing here between typical "Algebra I" and "Algebra II" - I think "I" is enough.
2023-03-03 12:19:22 <KipIngram   > Well, wait - maybe not.  I forget where multi-variable algebra is covered; that may be II.  And if you go beyond 2D you do start to pick up multi-variable stuff.
2023-03-03 12:54:14 <KipIngram   > Aha - looks like "Intel Turbo Boost" was causing my tests to show a small amount of cross-test interference.
2023-03-03 12:54:45 <KipIngram   > I did some tests the other day - created two hours of very smooth traffic on a drive, ran that with nothing else running, and then repeated the same test with it also running on a second drive.
2023-03-03 12:54:57 <KipIngram   > I could see just a small amount of "extra wiggle" in the 2-up test.
2023-03-03 12:55:27 <KipIngram   > But the SPDK developers suggested that I turn off Turbo Boost.  I did that and repeated the 2-up test this morning, and sure enough that extra wiggle vanished.
2023-03-03 12:56:03 <KipIngram   > If I understand what I read about Turbo Boost, it lets the processor move sections of code around from slower cpus to faster cpus, independent of my cpu pinning specifications.
2023-03-03 12:56:19 <KipIngram   > I think by turning it off I deny the processor that chance, and it properly respects my core assignments.
2023-03-03 12:56:30 <KipIngram   > Next work will be to pile on more drives; 3-up, 4-up, etc.
2023-03-03 12:56:41 <KipIngram   > See how far I can push this baby.
2023-03-03 12:57:16 <KipIngram   > In our actual product we probably want it on - anything that results in more across-the-board work getting done is desirable.
2023-03-03 12:57:23 <KipIngram   > But in my case I want *isolation*.
2023-03-03 12:57:48 <KipIngram   > The extra wiggle was quite minor to begin with, but I can't see it at all with turbo boost off.
2023-03-03 15:29:33 <Zarutian_iPad> a small project for anyone here new to forth, port https://github.com/google/open-location-code/blob/main/js/src/openlocationcode.js aka plus code decoder encoder to forth