3 ms·

You're actively ignoring the point and just being offended which is natural because it was put extremely bluntly. But you've just restated that the current stat

You're actively ignoring the point and just being offended which is natural because it was put extremely bluntly. But you've just restated that the current state of affairs is "entirely reasonable" when the comment you are replying to explains why it's actively harmful. You've provided no argument for why it's reasonable, just taken offense and waved around cases of catastrophic failure.

dekhn 8y ago
I don't understand what's unreasonble about: 1) companies have a profit motive and they want to design and sell products 2) historically, we have seen that companies acting with profit motives often have incentives to ignore problems with their products and even actively suppress them 3) the government wants to incentivize companies to make new products that help people 4) but they also don't want large numbers of people to die, or even small numbers of people, due to errors of any kind. 5) we have ample evidence that when unregulated, errors occur and people die. Statistically across large numbers of drugs and medical treatments, we have enough data to make strong, but not perfect statements about which processes minimize negative problems while maximizing availability of product to sick people. In some cases, people die because executives explicitly chose to cut corners. With regulation, we can catch those executives and put them in jail. 6) in general, the one thing that comes out of all of this is that extensive record keeping is required for investigators to deal with problems when they occur. Most of regulation is just record keeping- what did we do, what happened when we did it- and change control. Literally, my job was to ensure that firmware updates didn't change the observable properties of a system that was making a drug that saved people. Yes, when a firmware of a machine that makes drugs is updated, and it has a bug, people can die. 7) given about a hundred years of experience in this area, enough people who work on "reliability" (like SRE in tech) people have developed a set of techniques, called regulation, which attempts to minimize the number of deaths, while still allowing products to reach the market. Like SRE, it's a combination of hard-won experience and a lot of intelligence, along with not enough monitoring data. 8) I also feel pain and unhappiness that pharma doesn't move faster. It wasn't until I worked daily on product development and implementing regulation that I began to appreciate that this is a truly complicated problem with a lot of variables, incentives and players. 9) Even giving my best good faith effort, I simply cannot credit people who give anecdotal evidence about friends with diabetes. There are a number of reasons for this, but basically, if you do enough large-scale decision making, you'll find that anecdotes introduce bias that can lead to poor decision making. I see problems, also, in the DIY community where people sort of assume the best possible results; in science, after you run enough experiments, you get to see: it's easy to fool yourself and think what you've done is good, but it really has to percolate through the system and get evaluated and reproduced by many people before you can trust it. If you don't think the above logic chain is reasonable, that's OK. Many people aren't happy with how slowly pharma and medical DX industries and FDA work. But, you should basically understand that the entire goal here is to ensure good products make it to the market without "too many unnecessary deaths". The worst part of all of this is that I dedicated many years of my life to improving the speed at which drugs could be discovered and brought to patients, only to learn that the primary reasons we can't, are political, structural, systemic, and incentivized. If somebody had a better solution than what the current market has, I can point funding at them.

dekhn 8y ago

I don't understand what's unreasonble about: 1) companies have a profit motive and they want to design and sell products 2) historically, we have seen that companies acting with profit motives often have incentives to ignore problems with their products and even actively suppress them 3) the government wants to incentivize companies to make new products that help people 4) but they also don't want large numbers of people to die, or even small numbers of people, due to errors of any kind. 5) we have ample evidence that when unregulated, errors occur and people die. Statistically across large numbers of drugs and medical treatments, we have enough data to make strong, but not perfect statements about which processes minimize negative problems while maximizing availability of product to sick people. In some cases, people die because executives explicitly chose to cut corners. With regulation, we can catch those executives and put them in jail. 6) in general, the one thing that comes out of all of this is that extensive record keeping is required for investigators to deal with problems when they occur. Most of regulation is just record keeping- what did we do, what happened when we did it- and change control. Literally, my job was to ensure that firmware updates didn't change the observable properties of a system that was making a drug that saved people. Yes, when a firmware of a machine that makes drugs is updated, and it has a bug, people can die. 7) given about a hundred years of experience in this area, enough people who work on "reliability" (like SRE in tech) people have developed a set of techniques, called regulation, which attempts to minimize the number of deaths, while still allowing products to reach the market. Like SRE, it's a combination of hard-won experience and a lot of intelligence, along with not enough monitoring data. 8) I also feel pain and unhappiness that pharma doesn't move faster. It wasn't until I worked daily on product development and implementing regulation that I began to appreciate that this is a truly complicated problem with a lot of variables, incentives and players. 9) Even giving my best good faith effort, I simply cannot credit people who give anecdotal evidence about friends with diabetes. There are a number of reasons for this, but basically, if you do enough large-scale decision making, you'll find that anecdotes introduce bias that can lead to poor decision making. I see problems, also, in the DIY community where people sort of assume the best possible results; in science, after you run enough experiments, you get to see: it's easy to fool yourself and think what you've done is good, but it really has to percolate through the system and get evaluated and reproduced by many people before you can trust it. If you don't think the above logic chain is reasonable, that's OK. Many people aren't happy with how slowly pharma and medical DX industries and FDA work. But, you should basically understand that the entire goal here is to ensure good products make it to the market without "too many unnecessary deaths". The worst part of all of this is that I dedicated many years of my life to improving the speed at which drugs could be discovered and brought to patients, only to learn that the primary reasons we can't, are political, structural, systemic, and incentivized. If somebody had a better solution than what the current market has, I can point funding at them.