Author
|
Topic: Nutrition (Read 14134 times)
|
|
|
|
|
Julie.chan
*Many bubbles*
 Offline
Posts: 130
Sharing is good.
|
What do you think I am proposing? For a government agency to collect information about what you are buying? No, I am proposing that information that is already collected about what you are buying is sent to YOUR OWN account, not to a government agency.
There are so many things wrong with this I don't even know where to begin.
First of all, what you are describing now is just a form of self-reporting. We already do that.
So I presume what you're trying to do is prevent people from misremembering or misreporting data by forcing them to store it in a server somewhere -- you propose that only they can access it -- and then either helping or forcing them to then send their data to a research center.
For starters, your premise, that it is possible to force people to send data to a centralized server they don't control, using another computer they don't control, without anyone gaining access to your data, is simply untrue. Forget encryption or any other security measures. You're inputting this data into a store's POS system, which then encrypts it for you and sends it off to the government data server. If the government wants to track you, all they need to do is compel the store to steal your secret key as you type it in; the computer that handles your key entry can simply first store it in a text file somewhere and then use it to encrypt your data as normal.
This is of course assuming your key isn't laughably easy to guess, which if you know anything about people, you know is typically not the case. So many people choose obvious PINs like "1111", "1234", or "2468". In fact, if it's a 4-digit number, or even an 8 digit number, it's going to be easy to crack even if it's a "good" one. For any password to actually work against brute-force attacks, you need to incorporate both letters and numbers at least, and preferably special characters as well. That would mean you would have to install full-sized keyboards at every POS terminal for customers to use, or more realistically (since there's no way companies are going to go that far), use touchscreen keyboards that a decent camera can see very easily.
The only possible way I could imagine this problem being alleviated is if you never, ever identify yourself when using your key (which we have established you are not in favor of; you consider cash to be a relic of the "stone age", which is odd since cash didn't exist at that time), and you make it so that people set up their own account, again without identifying themselves, at their own discretion. Well, here's the problem with that: if anyone can set up a new account any time they like, people are going to be forgetting their keys, rebelling against the logging, etc. So a whole bunch of datasets are going to be put out of whack.
Plus, what ever happened to only collecting data about single people? You can't do that if you can't identify which data set corresponds with which person. And you have to know enough about each person to determine whether or not they are single. That sure sounds like a surveillance state to me in and of itself.
By the way, you seem to be really fond of laws preventing access. Those don't work. The government itself is your adversary in this case. Governments, as I said, can either weasel their way around laws, or pass new laws. You cannot depend on laws to stop your data from being looked at. That's why you need to stop your data from being collected in the first place.
I'm sorry, but you're trying to join two things that are diametrically opposed. There are two options:
1. Rely on self-reporting. This is what we currently do, and it works in any type of society, but it is often imperfect.
2. Use the power of a totalitarian dictatorship to collect the data of every citizen, but be more effective at doing so than North Korea. This would be perfect data as long as the government doesn't ever lie.
And a final point, even with option 2, it is not correct to say that this is equivalent to a randomized controlled trial. Research based on data from option 2 is still just a case-control study or other observational study.
|
|
|
|
|
Logged
|
|
|
|
Zanthius
Enlightened
Offline
Posts: 941
|
First of all, what you are describing now is just a form of self-reporting. We already do that.
Ehh.. No way. Even if it is your account, and only you have access to it, you should of course not be allowed to edit the data in any way. So, if you want to use this data for research, it is exactly the same data that already has been stored about you other places.
For starters, your premise, that it is possible to force people to send data to a centralized server they don't control, using another computer they don't control, without anyone gaining access to your data, is simply untrue.
Yeah, but since this account is only going to store information that already is stored about you somewhere else on the Internet, a hacker or government agency could just as well go there to get it. Probably easier to have supervision of the security in this server, than all the sources it is collecting data from.
Also, the server can be based on Linux and use a open source security system, and if we find out that someone has stolen information from the server, the population will of course demand an even more secure system.
Plus, what ever happened to only collecting data about single people? You can't do that if you can't identify which data set corresponds with which person.
Ehhh.. the information about if the person is single or not, is stored in the anonymous data set....
And a final point, even with option 2, it is not correct to say that this is equivalent to a randomized controlled trial.
Of course it is not equivalent to a randomized controlled trial. A randomized controlled trial is equivalent to one of your election polls. This is similar to a real election, since it is based on the entire population.
|
|
|
|
« Last Edit: July 04, 2017, 04:12:05 pm by Zanthius »
|
Logged
|
|
|
|
Julie.chan
*Many bubbles*
Offline
Posts: 130
Sharing is good.
|
Even if it is your account, and only you have access to it, you should of course not be allowed to edit the data in any way.
Then do you accept that what you are proposing would require all citizens to identify themselves, and for the government to be able to identify any particular person's purchase history?
Yeah, but since this account is only going to store information that already is stored about you somewhere else on the Internet, a hacker or government agency could just as well go there to get it.
If this information is already on the Internet, why don't researchers just use that? If they do use it, then why do you need to collect data in the first place?
A randomized controlled trial is equivalent to one of your election polls. This is similar to a real election, since it is based on the entire population.
Then you think observational studies are more reliable than randomized controlled trials? Do you have any evidence for this? It is, after all, the exact opposite of the current common understanding of scientific evidence, as you can see on the Wikipedia page for randomized controlled trials.
|
|
|
|
|
Logged
|
|
|
|
Zanthius
Enlightened
Offline
Posts: 941
|
Then you think observational studies are more reliable than randomized controlled trials? Do you have any evidence for this? It is, after all, the exact opposite of the current common understanding of scientific evidence, as you can see on the Wikipedia page for randomized controlled trials.
Look. This is basic set theory (mathematics). You agree that both the control group, and the treatment group in a randomized controlled trial are subsets of the entire population?
If you agree that both the control group and the treatment group are subsets of the entire population, do you also agree that all the information contained in those 2 subsets also is contained in the set of the entire population?
The reason why they are using a random selection of people in randomized controlled trials, is because they want to estimate the general population, since they don't have capacity to measure the entire population:
Since I want to study the entire population, I will just use the entire population that isn't getting the treatment as a control group:
|
|
|
|
« Last Edit: July 04, 2017, 05:20:45 pm by Zanthius »
|
Logged
|
|
|
|
Julie.chan
*Many bubbles*
Offline
Posts: 130
Sharing is good.
|
If you agree that both the control group and the treatment group are subsets of the entire population, do you also agree that all the information contained in those 2 subsets also is contained in the set of the entire population?
Yes, but this isn't a question about how much information you have. It's a question of methodology. If you take two groups of 20 rats, control every variable about both groups so that they are as identical as possible, then inject one group with cyanide, when you observe that 19 out of the 20 rats from the cyanide group died while none in the control group died, you can be confident that it was the cyanide that caused them to die and not the food you were feeding them. But if you have a million rats and somehow figure out exactly what all of them are eating, then correlate a high consumption of grass with a high rate of testicular cancer, you cannot confidently conclude that grass causes testicular cancer. This is basic science.
Since I want to study the entire population, I will just use the entire population that isn't getting the treatment as a control group
You're not performing an experiment, you're observing the current state of things. An actual experiment (a randomized controlled trial) would be to divide rats into two groups, separate them, control all variables you can (that's the controlled part of "randomized controlled trial"), and specifically add grass to one of the rat groups' diet. Then you check to see if that group has a higher rate of testicular cancer than the other group, and if the difference is statistically significant. If so, then you can conclude that grass causes testicular cancer in rats. Of course, the more such experiments show such a result, the more confident you can be that grass causes testicular cancer.
I would still like to know the answers to the two questions I posed in that same post, by the way:
Even if it is your account, and only you have access to it, you should of course not be allowed to edit the data in any way. Then do you accept that what you are proposing would require all citizens to identify themselves, and for the government to be able to identify any particular person's purchase history? Yeah, but since this account is only going to store information that already is stored about you somewhere else on the Internet, a hacker or government agency could just as well go there to get it. If this information is already on the Internet, why don't researchers just use that? If they do use it, then why do you need to collect data in the first place? |
|
|
|
« Last Edit: July 04, 2017, 05:25:09 pm by Julie.chan »
|
Logged
|
|
|
|
Zanthius
Enlightened
Offline
Posts: 941
|
Yes, but this isn't a question about how much information you have. It's a question of methodology. If you take two groups of 20 rats, control every variable about both groups so that they are as identical as possible, then inject one group with cyanide, when you observe that 19 out of the 20 rats from the cyanide group died while none in the control group died, you can be confident that it was the cyanide that caused them to die and not the food you were feeding them.
You would just use all the rats that didn't get cyanide as a control group, and you would still see that the death rate was much higher for the rats that got cyanide than for the general population of rats.
And btw, maybe you can control all the variables for a population of rats, but for ethical reasons they never do that in nutritional or medicinal studies on humans. You can just forget about controlling all the variables when studying humans. I try to control all the variables sometimes when I do chemical experiments, and even that is hard.
But yeah, I agree with you, that if we could control all the variables for a bunch of humans, that would also be a good way to study humans. But then you would need to treat those humans like research rats.
If this information is already on the Internet, why don't researchers just use that? If they do use it, then why do you need to collect data in the first place?
This is kinda what we discussed much earlier in the thread. It is better if data about me is sent to me, than if somebody collects data about me behind my back.
Then do you accept that what you are proposing would require all citizens to identify themselves, and for the government to be able to identify any particular person's purchase history?
I don't understand your question.
|
|
|
|
« Last Edit: July 04, 2017, 06:02:11 pm by Zanthius »
|
Logged
|
|
|
|
Julie.chan
*Many bubbles*
Offline
Posts: 130
Sharing is good.
|
You would just use all the rats that didn't get cyanide as a control group
The cyanide injection example was an example of an RCT. It was not an example of a case-control study. I don't know why you are suggesting using a different, less reliable control group (less reliable because you aren't controlling it for factors) in an example where a control group already exists.
And btw, maybe you can control all the variables for a population of rats, but for ethical reasons they never do that in nutritional or medicinal studies on humans. You can just forget about controlling all the variables when studying humans.
That's exactly what I said at the very beginning. Don't you remember? You made the argument that nutrition could be properly understood if only we gathered all of the data. Then you used North Korea as an example of a country that could surely advance nutrition science so far just by observing what it knows about its population. I objected, saying that I am not okay with giving up fundamental liberties for what I suppose would be an incremental improvement in our understanding at best. You agreed that you have no evidence of the extent to which monitoring people's purchases would improve scientific understanding of nutrition. So what we are left with is that we are looking at trading essential liberties for possibly improving scientific understanding by an undetermined amount. I do not think that is a good deal.
Your current argument, as I understand it, is that:
1. The data is already out there, so you don't need to worry about a loss of privacy.
2. Really big case-control studies are more reliable than randomized controlled trials.
So on the first point, I asked why researchers don't use that, which you have answered. On the second point, you are expressing a view that is diametrically opposed to the common understanding of the weight of different kinds of research, so for that, I have another question:
Do you have any evidence that sufficiently large case control studies are more reliable than properly sized randomized controlled trials?
I don't understand your question.
The question referred to is, "Do you accept that what you are proposing would require all citizens to identify themselves, and for the government to be able to identify any particular person's purchase history?"
What part of the question are you having trouble understanding?
It is better if data is sent to me, than if somebody collects data about me behind my back.
Are you saying that researchers don't use this data because of privacy concerns?
|
|
|
|
« Last Edit: July 04, 2017, 06:08:59 pm by Julie.chan »
|
Logged
|
|
|
|
Zanthius
Enlightened
Offline
Posts: 941
|
Do you have any evidence that sufficiently large case control studies are more reliable than properly sized randomized controlled trials?
Do you mean better than randomized controlled trials on humans, or on rats where you control all the variables? We never control all the variables in studies on humans.
Btw, I got this idea while doing chemistry experiments where I controlled all the variables, and only varied one variable. I imagined that if I just had billions of experiments, I could select experiments that only varied in one variable from chance, without necessarily setting up the experiment like that, and it would give me the same information.
You see... lets say there are 3 variables.
I could set up an experiment where I want to find the difference between:
[1][1][1] and [1][1][2]
But if I just had billions of experiments, I wouldn't need to set the first 2 variables to be equal, there would be some experiments like that from chance, and if I managed to find those experiments, it would be exactly the same.
Or to take another example. Lets say you wanted to study a vegetarian diet. You could set up an experiment where some people had to eat meat, while others had to eat vegetarian. But you could also just find carnivores and vegetarians in the general population, without setting up the experiment like that. When we have really big amounts of data, we can sometimes find examples that match the criteria we wanted to have for our controlled experiment. If you analyze the entire American population, there probably are a few individuals that eat almost exactly the same as you, except for differing in just one variable. Studying those individuals, and comparing them to people that eat more or less exactly the same as you, would be like a controlled experiment where you vary only one variable.
Do you accept that what you are proposing would require all citizens to identify themselves,
Of course, they would need to identify themselves to log into their accounts.
and for the government to be able to identify any particular person's purchase history?
I am not sure how the government is supposed to be able to this? By hacking into the server?
Are you saying that researchers don't use this data because of privacy concerns?
Yes, and because it is illegal for researchers and government workers to hack into grocery shops and hospitals to steal data.
|
|
|
|
« Last Edit: July 04, 2017, 07:51:15 pm by Zanthius »
|
Logged
|
|
|
|
Julie.chan
*Many bubbles*
Offline
Posts: 130
Sharing is good.
|
Do you mean better than randomized controlled trials on humans, or on rats where you control all the variables? We never control all the variables in studies on humans.
Comparing like with like. It's a generalized question. But if you need a specific example, then compare an RCT with rats with a hypothetical case control study on rats.
I imagined that if I just had billions of experiments
That's a meta-analysis. It's completely different from looking at correlations of populations.
I am not sure how the government is supposed to be able to this? By hacking into the server?
There are several possible ways. The simplest way is for the government to simply compel the shop to store your key (e.g. password) in a text file before using it to encrypt the data, then give that key to them. Since you were required to identify yourself, the government (or the shop itself, for that matter) can then connect the two. And this only ever has to happen once, at which point your entire history of purchases anywhere is now visible to the government, as well as all future purchases.
You cannot forcibly collect data from someone and not have the means to learn it at the same time.
Yes, and because it is illegal for researchers and government workers to hack into grocery shops and hospitals to steal data.
Do you think researchers are justified in their aversion to violating people's privacy in this way?
If so, why would it be any better for a government to force everyone to identify themselves any time they make a purchase?
If not, why not?
|
|
|
|
|
Logged
|
|
|
|
Zanthius
Enlightened
Offline
Posts: 941
|
Comparing like with like. It's a generalized question. But if you need a specific example, then compare an RCT with rats with a hypothetical case control study on rats.
Why should we compare this to a study on rats? Yes, I agree with you that studies on rats can be at least as reliable as the system I am proposing for studying humans, since we can do whatever we want with rats. We can't do whatever we want with humans. We can't test for lethal dose on humans. Doing research on humans is DIFFERENT from doing research on rats, and the system I am proposing was never intended for studying rats. In fact, I don't necessarily even think we need a better method for studying rats. We need a better system for studying humans.
There are several possible ways. The simplest way is for the government to simply compel the shop to store your key (e.g. password) in a text file before using it to encrypt the data, then give that key to them. Since you were required to identify yourself, the government
(or the shop itself, for that matter) can then connect the two. And this only ever has to happen once, at which point your entire history of purchases anywhere is now visible to the government, as well as all future purchases.
What is your proposal? To make all private companies and government institutions stop storing data about you? Good luck with that. There is always a possibility that someone might steal data about you if it is stored somewhere. Right now information about you is stored in hospitals, grocery shops, banks, several government institutions, and probably on lots of websites (like google and facebook). All of them can be hacked, just like your own email account can be hacked. I don't understand why it would be any worse if I got information stored about me by private companies and government institutions sent to my private account. It can just as well be stolen from their accounts as from my private account. Also, I kinda want to know what they have stored about me.
If so, why would it be any better for a government to force everyone to identify themselves any time they make a purchase?
When have I ever proposed that we should force people to identify themselves when they are buying groceries? I have said that grocery shops already are storing information about what you are buying, without your knowledge, as long as you are paying with a bank card.
Actually, according to the article I posted here earlier, they even store information about you if you pay with cash. There are several ways they could do this. For example by facial recognition cameras.
https://www.theguardian.com/money/2013/jun/08/supermarkets-get-your-data
That's a meta-analysis. It's completely different from looking at correlations of populations.
Yeah!! That sounds right. What I am proposing is like meta-studies based upon thousands of case-control correlation studies.
|
|
|
|
« Last Edit: July 04, 2017, 09:13:12 pm by Zanthius »
|
Logged
|
|
|
|
Julie.chan
*Many bubbles*
Offline
Posts: 130
Sharing is good.
|
Why should we compare this to a study on rats?
You really need to work on your reading comprehension. You have been constantly misunderstanding things I have been saying, and none of what you have been misunderstanding has been unclear.
I did not say anything about comparing human studies to rat studies. I asked you if you have any evidence that large case control studies are more reliable than properly sized, but smaller randomized controlled trials. It has nothing to do with rats or humans or any other research topic for that matter. It's only about the scientific method.
The question came about because you implied the argument that more data equates to better research. I am under the impression that you believe that properly done case control studies are at least as reliable as randomized controlled trials. Correct me if I am wrong. Otherwise, I would like to know if you have any evidence to support this belief.
What is your proposal? To make all private companies and government institutions stop storing data about you? Good luck with that.
Actually, yes.
But in the meantime, it's important to not force companies to collect, because then at least they won't always collect. Today, it is possible to purchase things anonymously with cash and without identifying yourself. If the government were to force shops to identify you, or abolish anonymous forms of payment, it would not be possible.
In other words: keep the status quo for now, work toward improving it. I do not accept defeatism as an option.
When have I ever proposed that we should force people to identify themselves when they are buying groceries?
Right here:
Do you accept that what you are proposing would require all citizens to identify themselves, Of course, they would need to identify themselves to log into their accounts.
We previously established that you are proposing that it should be impossible for the citizens to "edit the data":
Even if it is your account, and only you have access to it, you should of course not be allowed to edit the data in any way.
So that implies that you are required by law to access the account and send accurate information to it; to do otherwise would be to edit the data. This would in turn require implementing some mechanism to ensure that you are using your own key (I suppose by checking a cryptographic signature), and that would be impossible without identifying yourself. The shop would also have to verify that the information you sent is correct.
Therefore, what you are proposing necessarily requires everyone to identify themselves every time they make a purchase. The easiest way to do what you are proposing is to handle all of this encryption and verification in the shop's computer.
|
|
|
|
« Last Edit: July 04, 2017, 10:09:30 pm by Julie.chan »
|
Logged
|
|
|
|
Zanthius
Enlightened
Offline
Posts: 941
|
The question came about because you implied the argument that more data equates to better research. I am under the impression that you believe that properly done case control studies are at least as reliable as randomized controlled trials. Correct me if I am wrong. Otherwise, I would like to know if you have any evidence to support this belief.
I can guarantee you that in the randomized controlled experiments on humans, they aren't controlling all the variables. Yes, this is about the scientific method. Unless you control all the variables in a randomized controlled experiment on humans, my system is better, since it can find individuals that match the criteria you need for an experiment where you control all the variables from an enormously large set of data. There are probably several individuals in the United States that eat almost exactly the same food as you, but you can't necessarily force a bunch of people to eat exactly the same as you in an experiment. That would be unethical.
So that implies that you are required by law to access the account and send accurate information to it; to do otherwise would be to edit the data. This would in turn require implementing some mechanism to ensure that you are using your own key (I suppose by checking a cryptographic signature), and that would be impossible without identifying yourself. The shop would also have to verify that the information you sent is correct.
Look, I am not proposing that it should be sent to my private computer, but more like a super secure gmail account which only I have access to. That server could very well be made so that individuals could log in and look at their own data, but they wouldn't have any possibility to edit their data. It could also be made so that in order to log into your account, you would need to verify your identity with a password and code sent to your phone. The company that made the server, should also make it open source, so that we can look into the source code and see that they themselves also don't have any possibility to look at our data. That puts an even higher level of security on the server than what most countries have in their military. Because the military in many countries (like Iran), apparently use Windows, which made it open to attack by the Stuxnet virus. If I had a country, I would never let my military use Windows and/or macOS since it isn't open source.
Today, it is possible to purchase things anonymously with cash and without identifying yourself. If the government were to force shops to identify you, or abolish anonymous forms of payment, it would not be possible.
Maybe you can get a bank card that is related to a bitcoin account, and I kinda agree with you that this should be a possibility, because I like that people should be free to choose. Look what I found:
 
But you should know that most people don't care nearly as much about their privacy as you might, and the vast majority of people are never going to start using bank cards related to anonymous bitcoin accounts. The vast majority of people are not going to start using VPN when they are connected to Internet. The vast majority of people are not even going to stop putting out their private stuff on Facebook, because the vast majority of people don't care a lot about their privacy.
|
|
|
|
« Last Edit: July 04, 2017, 11:05:12 pm by Zanthius »
|
Logged
|
|
|
|
Zanthius
Enlightened
Offline
Posts: 941
|
I just realized now, that you might have been talking about the randomization between those that get the real pill and the fake pill. Such a randomization is of course not an option in most nutritional studies, since it is more or less impossible to give people fake food, like you give them a fake pill in pharmaceutical studies.
Well, anyhow, since you guys are so much into empirical evidence, here is something I found about that type of randomization:
Authors' conclusions
Our results across all reviews (pooled ROR 1.08) are very similar to results reported by similarly conducted reviews. As such, we have reached similar conclusions; on average, there is little evidence for significant effect estimate differences between observational studies and RCTs, regardless of specific observational study design, heterogeneity, or inclusion of studies of pharmacological interventions. Factors other than study design per se need to be considered when exploring reasons for a lack of agreement between results of RCTs and observational studies. Our results underscore that it is important for review authors to consider not only study design, but the level of heterogeneity in meta-analyses of RCTs or observational studies. A better understanding of how these factors influence study effects might yield estimates reflective of true effectiveness.
https://www.ncbi.nlm.nih.gov/pubmed/24782322
Authors' conclusions
The results of randomised and non-randomised studies sometimes differed. In some instances non-randomised studies yielded larger estimates of effect and in other instances randomised trials yielded larger estimates of effect. The results of controlled trials with adequate and inadequate/unclear concealment of allocation sometimes differed. When differences occurred, most often trials with inadequate or unclear allocation concealment yielded larger estimates of effects relative to controlled trials with adequate allocation concealment. However, it is not generally possible to predict the magnitude, or even the direction, of possible selection biases and consequent distortions of treatment effects from studies with non-random allocation or controlled trials with inadequate or unclear allocation concealment.
https://www.ncbi.nlm.nih.gov/pubmed/17443633
What a sad, sad, state the sciences that study humans are in.... I can just imagine if I randomly assigned a "real" chemical and a "fake" chemical to my reactions, without having any control of all the other variables, and there would be thousands of other chemicals in these reactions, which I had no control over. I doubt it would be possible for me to figure out how that specific chemical influenced the reaction....
|
|
|
|
« Last Edit: July 05, 2017, 12:15:16 am by Zanthius »
|
Logged
|
|
|
|
|
