10,104 karma · joined July 1, 2009
https://monocypher.org
https://elligator.org
"My dad's a genius" because you're not supposed to rely on genius to make a good crypto system, and also because it makes Eap sounds like he has absolutely zero knowledge on the subject.
"highest level of encryption available" because there's a fairly low floor above which it's all uncrackable anyway (ChaCha20 + BLAKE2B authenticated encryption, and Curve448 + post quantum winners for the public stuff, should go beyond total overkill).
I don't believe it was intentional though. I'm just out of a quick job implementing SSCPv2 (encryption over RS485 to secure communication between card readers and central computer, typically used to secure buildings). Good specs, fairly good separation between cryptography and business logic, and as far as I could tell the crypto isn't broken… but it is quite old school: AES CBC + HMAC SHA256, using MAC then encrypt. https://moxie.org/2011/12/13/the-cryptographic-doom-principl... And while I think my implementation is okay, I did have to pay special attention to specific traps raising from this design, and to be honest wouldn't bet my life on having ironed out all possible timing attacks.
SSCPv2 was almost certainly designed after 2020, but it took books from 2005. Good books for their time, but a bit dated unfortunately. I'm pretty sure no actual cryptographer was involved. If there were, they would almost certainly have used standard authenticated encryption scheme like AES CGM, or ChaPoly (RFC 8439), they would have authenticated the unencrypted header, and provided an even better separation between crypto and business logic.
Quite a few companies have millions and millions of lines of code. Changing 1% of it would mean changing more than 10K lines of code, perhaps even more than 100K. In much bigger code bases, where changing anything has a risk of breaking something — not just because you might make a mistake, but because your program is full of Undefined Behaviour, and changing anything might manifest latent bugs.
Given that, I'm not surprised people say that Sutter quote with a straight face.
Does your unemployment insurance work after you quit? Or does it only work when you're fired?
In practice, the people who make the calls more often does not do so because they inherited a crap load of money, not because their own genius or hard work raised them to the top.
Wait a minute if the people most suited to make a decision are overridden by people less competent than them (they have to be most of the time, given the different focus of their career), that's kind of a problem, isn't it? Is there any way to avoid such structural failures?
I can.
Think of what employment is for a second: shareholders (or company owners) own the capital, and the employee gets to follow orders. Structurally, the company pay workers less than their actual value: shareholders gotta hold, and they squeeze the margin out of the employee. The margin may be thin, but it's never meant to be zero (except for non-profits, but they're the exception).
Such a relationship is fundamentally asymmetrical, such an exchange fundamentally unequal. If you want any hope of restoring fairness from this system, termination conditions have to be asymmetrical as well.
I would agree in an ideal world where salaries are tied to a qualification (a notion a tad more nuanced than just diploma), instead of their job. While there is no (nor should be any) constitutional right to work at any given company, I think people should have an inalienable right to live. Which in practice means food, shelter, and health care at a minimum, regardless of their ability (perhaps even willingness) to work anywhere.
The compromise most EU countries have settled on is that once you're employed and past some probation period (in my line of work that can last up to 8 months), then they can't fire you without a damn good reason or a hefty severance (the better the reason the lower the severance, basically). But it cuts both ways: I personally can't leave immediately, I have to tell my employer 3 months in advance. It is in a way a kind of severance.
Ok, cool, I've heard this a gazillion times by now. I don't disagree, it does looks like it's better at getting shit done. Just one question: how do I get credited for anything? I have my career to look out for too.
Possibly: in many places there they have "at will" employment, where they can basically fire you with little to no severance at any time for any reason. There are limits, but compared to most of Europe that's the gist of it.
So yeah, you can be fired just because someone doesn't like you. And giving feedback is a good way to not be liked.
The consequences of bad management in a software project on the other hand are much more nebulous. And nobody died as a direct consequence of some manager being a jerk… oh wait, I'm sure some did (committed suicide or gone postal), but it's easy to just blame the worker instead of the working place.
It seems the real problem here is the power difference between people, combined with the powerful side not having a thick enough skin… or lacking the actual competence that we ought to expect of their position.
A competent leader will not be offended or feel threatened by a piece of feedback, however negative: either the feedback is crap and they'll calmly say "nope, trust me I know what I'm doing here", or the feedback has enough truth in it for them to say "oops, I'll do better next time", and then proceed to actually do better.
An incompetent leader however likely know they're incompetent, and they're less likely to meaningfully judge feedback. They're more likely to look for social validation, and more importantly secure their position. Any negative feedback threatens that position, and when that feedback comes from below, the solution is obvious: slap the peasant down, perhaps even fire them.
Often there's little you can do around those people. For instance, I once applied for a contracting gig that would involve cryptographic work. I was rejected because I was "more competent than the project lead". My guess is the hiring manager there knew that if they brought in someone more competent than the project lead, things would go bad. Quite the indictment of the project lead if you ask me.
My best project lead on the other hand had no problem being a worse programmer than the people under him (the tech lead for sure, and I probably). He worked on the parts he could, trusted us to do our thing, and I trusted him with telling me the priorities. Best gig of my entire career.
Starting from there (to be checked), shortage of equipment is not a thing: most routers are programmable computers, all we need to do is update the software to something that doesn't suck. And best of all, the customer doesn't even have to know. They just wake up one morning, and as if by magic, their Zoom/Meets/Hangouts calls suddenly work better.
Now if most home routers are not updated by the ISP… my condolences to your country I guess.
Edit: OK, real question: how does home routing actually work in the US? Who is responsible for what equipment, and for the ISP issued equipment, are there any relevant remote update capabilities, and if so are ISPs actually competent at doing updates?
You're a bit light on justifying the burden of proof, though.
Chile's attempt at lukewarm socialism didn't fail from internal causes. Cuba didn't brought embargo on themselves. To name but two. Considering the USA has been at war for almost the entirety of it's history, there must be a couple more.
"Club" was a tame euphemism I only took from the comment I was replying to. I think "Mafia" would be more appropriate.
I'm pretty sure they won't. And for that reason alone… https://www.youtube.com/@TacticoolGirlfriend
It's a bit more complicated than that. Gender is a social construct, mostly determined by genes & genitalia. It's not quite enough to believe you're a woman, other people have to believe it too. Another issue at play is that there are far more "intersex" people (who have some characteristics of the opposite sex, sometimes to the point doctors don't quite know whether to list them as male or female), and from what I've heard trans people often (possibly generally) are "intersex" in a way that wasn't visible at birth. The idea of a female's brain in a male's body isn't that far fetched.
> They believe in censorship.
I believe this one is more popular in the far right (when in power) than in the far left (when in power)
> They believe in supporting and growing the military industrial complex.
Militarism sounds like it's more popular on the right. Though it can be more complicated: military backed imperialism can indeed support stuff like welfare at home.
---
Now the elephant in the room: last time I checked, democrats were firmly capitalists: they believe the means of production should be owned privately. Even if you exclude actual communism from acceptable discourse, they're fairly poor at public services and keeping inequality in check.
Sorry if that feels like a strawman, but I find the idea of using popularity to determining what counts as "far" stupid and dangerous.
This feels like a club you pay 2% for protection…
But you know what, even that can be gamed. I have gamed it once: I copied & pasted the definition of the struct, and accessed its member from a place that was supposed to only have access to an opaque pointer. It worked. (To be fair, this isn't the kind of stuff you do by mistake, unlike merely accessing a struct member.)
Anyway, I don’t care about what we should mean by this and that. What I do care about is whether inheritance is useful or not. And so far, I have seen no evidence that it is anything more than highly situational — meaning, doesn’t have a better alternative for almost everything.
The study I recall didn’t measure productivity, it counted bugs.
Anyway, I spent enough time on the subject to close the case. Until I stumble upon cogent evidence to the contrary, inheritance is not worth my time. I won’t use it, I will steer my colleagues away from it, and I will quit gigs that use it too heavily.
Yeah, one of the 3 big lies of computer programming: thinking that we should program around a model of the world. We should program around a model of the data that we need to transform to solve whatever problem we need to solve. Even in simulation software like games.
A car have a motor which have pistons. But why would I model that? To be honest I’m not sure OOP is appropriate even for CAD software here. Now maybe we could give the user the ability to make a mesh and call that "piston", an let them copy & paste that mesh to build a super-mesh they would call "motor", but this kind of composition is not quite OOP, and the appropriate way to program this stuff is probably not OOP either.