4 ms·

This is not well researched at all. Anything that depends on “if everyone acted intelligently and good will” is broken. If you’re interested in academia I stron

This is not well researched at all. Anything that depends on “if everyone acted intelligently and good will” is broken. If you’re interested in academia I strongly recommend you go back and look at designing systems that function in the face of incompetence and even adversaries.

PakG1 3y ago
You can have your opinion, but I have mine after reading lots of research papers. Obviously, there are ways that things can still be improved. But even the design of systems that function in the face of incompetence and even adversaries will face intractable problems of incompetent leadership and governance. It's one thing to say "this is how it should be done" and it's another to say that we've managed to get people to do it the way it should be done. No matter how much improvement we see on ideas for how it should be done, we're still no closer to solving the second part, despite much effort. In the end, there will always be incompetent leaders in charge at some point for some project somewhere, and more often than we'd prefer. You can lead a horse to water, you can't make it drink.
- kortilla 3y ago
  It’s not an opinion and you can cop out to lame “what if the government is hitler” arguments, but resilient systems are definitely an engineering/science/math problem. The entire field of cryptography wouldn’t even exist if the boundary of research ended at “good actors”.
  - Joeboy 3y ago
    Surely the field of cryptography relies on conscientious and competent actors developing solutions that are robust in the face of malicious actors. I am skeptical that there are software development practices that will allow me to hire a team of feckless incompetents and have them develop quality software. If you know of any I'm interested to hear about them.
    - kortilla 3y ago
      That’s not necessary. The broader system of society, communication, and the encouragement of development of open solutions to this problem means your team of incompetents just doesn’t even need to solve them.
  - PakG1 3y ago
    And yet we still have people out there trying to create their own cryptography when the golden rule is to not roll your own crypto. For whatever reason, best practices don't get followed 100% of the time, even if they exist. For cryptography, the situation is better than most other domains. I think we're having different conversations here. You seem to be having a technical conversation. I'm having a sociotechnical conversations within the context of organizations and their workers and managers. I'm seeing you discuss technical solutions to sociotechnical issues, which is not what I am discussing. Even when the technical ideas are perfect, organizations still need to implement the ideas. That implementation tends to not follow allegedly perfect specifications for many reasons. But you are entitled to your opinion and that's fine. We can agree to disagree, nothing wrong with that.
    - kortilla 3y ago
      > For whatever reason, best practices don't get followed 100% of the time, even if they exist. For cryptography, the situation is better than most other domains. That’s the point. Cryptography is significantly better precisely because the research effort has gone into systems that are hard for people to fuck up. Just look at the fight was to get everyone to agree that the model should be, “everything including the algorithm should be public, except for the key”. That’s a socioeconomic argument. > Even when the technical ideas are perfect, organizations still need to implement the ideas. That implementation tends to not follow allegedly perfect specifications for many reasons. And that’s why making safe systems where mistakes are protected against is a critical area of research. Rust is popular because it protects against whole classes of bugs, despite it being no faster than C/C++.
      - PakG1 3y ago
        You fail to address the issue that none of these resolve the issue that people still don't follow best practices. This is especially true the further one is from best practices. The reason why cryptography is significantly better is not because the best research has been done. It's because it's impossible for cryptography work to be done without the best people because it's that hard. Drop cryptography engineers into this Post Office mess in the middle of the implementation and they'd say screw it, you can't pay me enough to deal with this crap. They'll leave to go work for an organization that respects them enough to listen to them and still make it worth their while. You can talk about best practices all you want. You're not going to get most organizations to afford or convince the best people capable of following best practices to come work for them. And even if they did, those best people will leave before they can even change the technical culture. No effective person would put up with the insanity that exists in subpar organizations, many of which continue to exist in spite of their incompetence for many other reasons. You have no concept of working in the real world where people who suck exist. You talk like you've only ever worked with all-star Linus Torvalds types. Of course it's easy to do what you are recommending when you're working on the Linux kernel, for FAANG, startups with competent founders, etc. All those organizations are able to do what you recommend for reasons that many other organizations can't. You're a Xoogler working with startups. I get it. You're in that world. You have no idea how to fix an organization like the British Postal Office so that they will do IT competently.
Joeboy 3y ago
Are there software dev processes that operate well in the face of incompetent or malicious people? I can think of ways to mitigate the damage, but at the end of the day surely you need some competent and conscientious people on your project.
- kortilla 3y ago
  varying levels of incompetency can be solved by using safer but slower languages. Malicious is harder in the face of multiple colluding actors. Automated flagging of suspect code by AI to wider groups for review and sign off would drastically raise the bar (but obviously not solve the sneakiest of cases).

PakG1 3y ago

You can have your opinion, but I have mine after reading lots of research papers. Obviously, there are ways that things can still be improved. But even the design of systems that function in the face of incompetence and even adversaries will face intractable problems of incompetent leadership and governance. It's one thing to say "this is how it should be done" and it's another to say that we've managed to get people to do it the way it should be done. No matter how much improvement we see on ideas for how it should be done, we're still no closer to solving the second part, despite much effort. In the end, there will always be incompetent leaders in charge at some point for some project somewhere, and more often than we'd prefer. You can lead a horse to water, you can't make it drink.

kortilla 3y ago
It’s not an opinion and you can cop out to lame “what if the government is hitler” arguments, but resilient systems are definitely an engineering/science/math problem. The entire field of cryptography wouldn’t even exist if the boundary of research ended at “good actors”.
- Joeboy 3y ago
  Surely the field of cryptography relies on conscientious and competent actors developing solutions that are robust in the face of malicious actors. I am skeptical that there are software development practices that will allow me to hire a team of feckless incompetents and have them develop quality software. If you know of any I'm interested to hear about them.
  - kortilla 3y ago
    That’s not necessary. The broader system of society, communication, and the encouragement of development of open solutions to this problem means your team of incompetents just doesn’t even need to solve them.
- PakG1 3y ago
  And yet we still have people out there trying to create their own cryptography when the golden rule is to not roll your own crypto. For whatever reason, best practices don't get followed 100% of the time, even if they exist. For cryptography, the situation is better than most other domains. I think we're having different conversations here. You seem to be having a technical conversation. I'm having a sociotechnical conversations within the context of organizations and their workers and managers. I'm seeing you discuss technical solutions to sociotechnical issues, which is not what I am discussing. Even when the technical ideas are perfect, organizations still need to implement the ideas. That implementation tends to not follow allegedly perfect specifications for many reasons. But you are entitled to your opinion and that's fine. We can agree to disagree, nothing wrong with that.
  - kortilla 3y ago
    > For whatever reason, best practices don't get followed 100% of the time, even if they exist. For cryptography, the situation is better than most other domains. That’s the point. Cryptography is significantly better precisely because the research effort has gone into systems that are hard for people to fuck up. Just look at the fight was to get everyone to agree that the model should be, “everything including the algorithm should be public, except for the key”. That’s a socioeconomic argument. > Even when the technical ideas are perfect, organizations still need to implement the ideas. That implementation tends to not follow allegedly perfect specifications for many reasons. And that’s why making safe systems where mistakes are protected against is a critical area of research. Rust is popular because it protects against whole classes of bugs, despite it being no faster than C/C++.
    - PakG1 3y ago
      You fail to address the issue that none of these resolve the issue that people still don't follow best practices. This is especially true the further one is from best practices. The reason why cryptography is significantly better is not because the best research has been done. It's because it's impossible for cryptography work to be done without the best people because it's that hard. Drop cryptography engineers into this Post Office mess in the middle of the implementation and they'd say screw it, you can't pay me enough to deal with this crap. They'll leave to go work for an organization that respects them enough to listen to them and still make it worth their while. You can talk about best practices all you want. You're not going to get most organizations to afford or convince the best people capable of following best practices to come work for them. And even if they did, those best people will leave before they can even change the technical culture. No effective person would put up with the insanity that exists in subpar organizations, many of which continue to exist in spite of their incompetence for many other reasons. You have no concept of working in the real world where people who suck exist. You talk like you've only ever worked with all-star Linus Torvalds types. Of course it's easy to do what you are recommending when you're working on the Linux kernel, for FAANG, startups with competent founders, etc. All those organizations are able to do what you recommend for reasons that many other organizations can't. You're a Xoogler working with startups. I get it. You're in that world. You have no idea how to fix an organization like the British Postal Office so that they will do IT competently.

Joeboy 3y ago

Are there software dev processes that operate well in the face of incompetent or malicious people? I can think of ways to mitigate the damage, but at the end of the day surely you need some competent and conscientious people on your project.

kortilla 3y ago
varying levels of incompetency can be solved by using safer but slower languages. Malicious is harder in the face of multiple colluding actors. Automated flagging of suspect code by AI to wider groups for review and sign off would drastically raise the bar (but obviously not solve the sneakiest of cases).