6 ms·

> it's the equivalent of flying a plane you built yourself A great analogy because people die that way. I personally would never push code to another person’s

> it's the equivalent of flying a plane you built yourself A great analogy because people die that way. I personally would never push code to another person’s insulin pump (or advertise code as being used for an insulin pump) because I couldn’t live with the guilt if my bug got someone else killed.

jacquesm 10mo ago
I know people die that way (GA). But someone is working for the companies that make insulin pumps and they are not as a rule equally motivated so I would expect them to do worse, not better. And to the best of my knowledge none of the closed-loop people have died as a result of their work and they are very good at peer reviewing each others work to make sure it stays that way. And I'd trust my life to open source in such a setting long before I'd do it to closed source. At least I'd have a chance to see what the quality of the code is, which in the embedded space ranges from 'wow' all the way to 'no way they did that'.
- chii 10mo ago
  > I would expect them to do worse, not better. which is why lots of systems and processes (sometimes called red tape) exist to try and prevent the undesired outcome, and dont rely on the competency of a single person as the weak link!
  - firtoz 10mo ago
    There are more financial reasons to violate and cheat the red tape than there are incompetent open source hackers in the world.
horsawlarway 10mo ago
And yet someone IS pushing code to these devices. Every single one. So the question really becomes - Are these people working on their own pumps with open source more or less invested than the random programmers hired by a company that pretty clearly can't get details right around licensing, and is operating with a profit motive? More reckless as well? Perhaps. But at least motivated by the correct incentives.
- dullcrisp 10mo ago
  So flying in a plane you built yourself is in fact safer than flying commercial because the motivations line up. Got it.
  - mindslight 10mo ago
    Flying in a plane you built yourself is likely safer than flying in the same model of plane built by a company that assembled it for you using lowest-bid labor while making you sign a twenty page lawyer barf disclaiming liability.
    - amrocha 10mo ago
      You can’t honestly believe that or you wouldn’t be able to function in society.
      - mindslight 10mo ago
        My comment rests on the fact that the types of planes you can build yourself are completely different models than the fully assembled models from the likes of Boeing etc. I do agree that a kit 737, if such a thing existed, would be less safe than one off the line.
        amrocha 10mo ago
        I would still trust a cessna way more than any plane built or modified by a single person.
        jacquesm 10mo ago
        I think the Beechcraft Bonanza deserves special mention here. I'm sure all the people that worked on it were experts too! The big problem with this analogy is that it conflates three very important things: - GA is more dangerous, period. Doesn't matter whether you build the plane yourself or if you bought it ready made (hopefully new, hopefully very well maintained if second hand) - GA craft tend to have less experienced pilots than airliners, but even airliner pilots tend to do worse as GA pilots than when they're at work. The reason for that is simple: the processes are what keeps commercial aviation (mostly) safe. - GA craft tend to kill the pilots, because they are more often than not the only person on the plane. - GA craft have malfunctions like larger aircraft, there is nothing special about them in that sense. But there is something that they don't have that larger aircraft do have: redundancy. In electronics systems, in the design of the mechanical bits, and finally in the people. - GA craft that are designed and built by their operators are experimental class for a reason: they are untested and so more likely to fail than the ones that are certified. The design processes for commercial aircraft are nothing compared to the design processes employed by what we'll call hobbyists to distinguish them. - And finally, even though it is a fun analogy I only meant it from a skin-in-the-game point of view, a GA hobbyist is still going to do his level best to make sure that he's not going to get killed. Boeing executives only care about the bottom line, safety is a distant second. And based on my experience with the difference between the guts of various bits and pieces of avionics and the software that they run on compared to my experience looking at medical devices, their guts and the software that they run on I would be more than happy to bet that the loop hackers know as much more more about the failure modes of these devices as the manufacturers do. Cleanroom manufacturing under sterile conditions is the main differentiator here, and that just applies to the hardware, and it is an art that the medical industry understands very well. Electronics is already at a lower level of competence and their software knowledge tends to be terrible, not to mention the QA processes on said software. Programmers working for corporations don't necessarily suddenly grow an extra quality brain when they do their work.
f1shy 10mo ago
Anytime anybody does something himself, there is a risk. People die because of welding parts cleaned with break-cleaner, people die driving, diving, sky-diving, doing bungee jumping... Advertising that code, IMHO would be as showing of you doing extreme sports, for example. I do not think is any bad. A good disclaimer should be enough to take away any guilt.
RobotToaster 10mo ago
I'm not aware of any deaths attributed to open source artificial pancreas systems. Meanwhile there have been multiple attributed to closed source glucose monitors.
- croemer 10mo ago
  Not attributed to. The FDA wording says "associated with" which is much weaker causally.
  - UomoNeroNero 10mo ago
    I can guarantee you, from my personal experience of being diabetic for 30 years, that every day—and in the most incredible ways—I have managed to “almost kill myself.” Whether when I used finger-prick testing, sensors, injecting insulin with pens, or managing insulin with a pump. Our life is always a delicate balancing act between too little, too much, and way too much—the kind where this time I really kick the bucket By personal choice I use a commercial CGM (if I could “touch it,” I’d be firmly on the side of certainty about killing myself through sheer stupidity), but reading something like “associated with” really makes me angry. Before making such subtle insinuations about the open-source world (the source of the revolution of the last 10 years in this field), regulatory bodies should open their eyes to what is actually happening with the quality of current sensors and the real problems they are causing.
    - jacquesm 10mo ago
      Thank you. And strength to you. I had a business partner for some time that was much like you and every time he'd be 10 minutes late for an appointment I'd get nervous and if it was more than an hour I'd be on the phone to his family to check up on him.
fsflover 10mo ago
https://news.ycombinator.com/item?id=46398414 https://news.ycombinator.com/item?id=46398414

jacquesm 10mo ago

I know people die that way (GA). But someone is working for the companies that make insulin pumps and they are not as a rule equally motivated so I would expect them to do worse, not better. And to the best of my knowledge none of the closed-loop people have died as a result of their work and they are very good at peer reviewing each others work to make sure it stays that way. And I'd trust my life to open source in such a setting long before I'd do it to closed source. At least I'd have a chance to see what the quality of the code is, which in the embedded space ranges from 'wow' all the way to 'no way they did that'.

chii 10mo ago
> I would expect them to do worse, not better. which is why lots of systems and processes (sometimes called red tape) exist to try and prevent the undesired outcome, and dont rely on the competency of a single person as the weak link!
- firtoz 10mo ago
  There are more financial reasons to violate and cheat the red tape than there are incompetent open source hackers in the world.

horsawlarway 10mo ago

And yet someone IS pushing code to these devices. Every single one. So the question really becomes - Are these people working on their own pumps with open source more or less invested than the random programmers hired by a company that pretty clearly can't get details right around licensing, and is operating with a profit motive? More reckless as well? Perhaps. But at least motivated by the correct incentives.

dullcrisp 10mo ago
So flying in a plane you built yourself is in fact safer than flying commercial because the motivations line up. Got it.
- mindslight 10mo ago
  Flying in a plane you built yourself is likely safer than flying in the same model of plane built by a company that assembled it for you using lowest-bid labor while making you sign a twenty page lawyer barf disclaiming liability.
  - amrocha 10mo ago
    You can’t honestly believe that or you wouldn’t be able to function in society.
    - mindslight 10mo ago
      My comment rests on the fact that the types of planes you can build yourself are completely different models than the fully assembled models from the likes of Boeing etc. I do agree that a kit 737, if such a thing existed, would be less safe than one off the line.
      - amrocha 10mo ago
        I would still trust a cessna way more than any plane built or modified by a single person.
        jacquesm 10mo ago
        I think the Beechcraft Bonanza deserves special mention here. I'm sure all the people that worked on it were experts too! The big problem with this analogy is that it conflates three very important things: - GA is more dangerous, period. Doesn't matter whether you build the plane yourself or if you bought it ready made (hopefully new, hopefully very well maintained if second hand) - GA craft tend to have less experienced pilots than airliners, but even airliner pilots tend to do worse as GA pilots than when they're at work. The reason for that is simple: the processes are what keeps commercial aviation (mostly) safe. - GA craft tend to kill the pilots, because they are more often than not the only person on the plane. - GA craft have malfunctions like larger aircraft, there is nothing special about them in that sense. But there is something that they don't have that larger aircraft do have: redundancy. In electronics systems, in the design of the mechanical bits, and finally in the people. - GA craft that are designed and built by their operators are experimental class for a reason: they are untested and so more likely to fail than the ones that are certified. The design processes for commercial aircraft are nothing compared to the design processes employed by what we'll call hobbyists to distinguish them. - And finally, even though it is a fun analogy I only meant it from a skin-in-the-game point of view, a GA hobbyist is still going to do his level best to make sure that he's not going to get killed. Boeing executives only care about the bottom line, safety is a distant second. And based on my experience with the difference between the guts of various bits and pieces of avionics and the software that they run on compared to my experience looking at medical devices, their guts and the software that they run on I would be more than happy to bet that the loop hackers know as much more more about the failure modes of these devices as the manufacturers do. Cleanroom manufacturing under sterile conditions is the main differentiator here, and that just applies to the hardware, and it is an art that the medical industry understands very well. Electronics is already at a lower level of competence and their software knowledge tends to be terrible, not to mention the QA processes on said software. Programmers working for corporations don't necessarily suddenly grow an extra quality brain when they do their work.

f1shy 10mo ago

Anytime anybody does something himself, there is a risk. People die because of welding parts cleaned with break-cleaner, people die driving, diving, sky-diving, doing bungee jumping... Advertising that code, IMHO would be as showing of you doing extreme sports, for example. I do not think is any bad. A good disclaimer should be enough to take away any guilt.

RobotToaster 10mo ago

I'm not aware of any deaths attributed to open source artificial pancreas systems. Meanwhile there have been multiple attributed to closed source glucose monitors.

croemer 10mo ago
Not attributed to. The FDA wording says "associated with" which is much weaker causally.
- UomoNeroNero 10mo ago
  I can guarantee you, from my personal experience of being diabetic for 30 years, that every day—and in the most incredible ways—I have managed to “almost kill myself.” Whether when I used finger-prick testing, sensors, injecting insulin with pens, or managing insulin with a pump. Our life is always a delicate balancing act between too little, too much, and way too much—the kind where this time I really kick the bucket By personal choice I use a commercial CGM (if I could “touch it,” I’d be firmly on the side of certainty about killing myself through sheer stupidity), but reading something like “associated with” really makes me angry. Before making such subtle insinuations about the open-source world (the source of the revolution of the last 10 years in this field), regulatory bodies should open their eyes to what is actually happening with the quality of current sensors and the real problems they are causing.
  - jacquesm 10mo ago
    Thank you. And strength to you. I had a business partner for some time that was much like you and every time he'd be 10 minutes late for an appointment I'd get nervous and if it was more than an hour I'd be on the phone to his family to check up on him.

fsflover 10mo ago

https://news.ycombinator.com/item?id=46398414 https://news.ycombinator.com/item?id=46398414