Author
|
Topic: Nutrition (Read 13652 times)
|
Scalare
*Many bubbles*
 Offline
Posts: 245
|
At work we also make a graphical model for every explanation. That's why people call us jokingly a model agency  .
|
|
|
|
|
Logged
|
|
|
|
Death 999
Global Moderator
Enlightened
Offline
Gender: 
Posts: 3874
We did. You did. Yes we can. No.
|
So what if the same argument has been used against Bernie? Even if it was invalid against him, that doesn't mean that it's unsound in general.
It's really hard to centrally aggregate information in such a fashion that individual trends can be extracted, without identifying the individual.
It would be really easy if we only needed the individual to be able to get their data, and it would be reasonably easy if only the individual needed to add their own information. But for lots of people to put in and get out information about that person, without being able to connect those two ideas? That seems… hard.
|
|
|
|
|
Logged
|
|
|
|
Zanthius
Enlightened
Offline
Posts: 941
|
So what if the same argument has been used against Bernie? Even if it was invalid against him, that doesn't mean that it's unsound in general.
It seems like a destructive negative-sum activity to argue in that way, and reminds me of what she accused Hillary of doing during the election:
I think the reason many people voted for Trump (or more accurately, the reason many people didn't vote for Hillary Clinton) was because Hillary Clinton completely failed to excite the Democratic base. Further, Trump supporters were heavily demonized, which made them afraid to speak in support of Trump, that led to severe overconfidence in Hillary's ability to win (the press essentially described her victory as inevitable, as if it was literally impossible for Trump to win), and so people just didn't bother voting.
Trump, on the other hand, although he was always a prolific liar, was able to project a message to his base about change and jobs and whatnot, and that excited his base. Meanwhile, his complaints of HIllary Clinton were about actual problems, while Clinton's campaign just focused on name-calling. And since voting is anonymous, the name-calling and attempt to silence Trump supporters not only didn't help Clinton, it made Trump supporters want to support Trump even more. Even then, Donald Trumps performance was not any better than John McCain or Mitt Romney. Clinton just performed much worse than Barack Obama.
Pinpointing exactly what is wrong with a theory, seems much more constructive than name-calling.
It would be really easy if we only needed the individual to be able to get their data, and it would be reasonably easy if only the individual needed to add their own information.
Something like this maybe?
|
|
|
|
« Last Edit: June 28, 2017, 07:06:14 pm by Zanthius »
|
Logged
|
|
|
|
|
|
Scalare
*Many bubbles*
Offline
Posts: 245
|
also, to be honest, like with cancer and it being a disease unique to you, it's better to focus on individuals themselves and have a neural net analyse info based on just one person instead of the general populace. If you hookup a personal AI to all your personal data it could figure out for yourself how to live most healthy, taking into account your personal body composition and digestive tract.
|
|
|
|
|
Logged
|
|
|
|
|
|
Scalare
*Many bubbles*
Offline
Posts: 245
|
Ie the grocery bills will not be sent to banks and to the evil government but directly from yourself to the researchers (why even bother with federal people involved).
Objection noted. I have renamed it from "Federal Statistics Bureau" to "Research Team" now, since I really don't care if it is federal or private, as long as it is anonymous. I have put all of it together on this page now: http://archania.org/a_new_way_to_do_nutritional_research.htmlThank you to all of you that have given me constructive negative feedback, which has helped me to develop this theory. Since I am apparently not getting any more negative feedback, I am assuming that everything is alright.
no, i'm just tired of pointing out everything that I might dislike 
|
|
|
|
|
Logged
|
|
|
|
Zanthius
Enlightened
Offline
Posts: 941
|
I haven't written about it, but of course this research system could also give us much more information about medicines, and if they interact with different types of food.
And I wouldn't exactly be surprised if lots of the medicines work much less than they tell us, and have a lot more side effects that we aren't aware of.
There. I have added pharmacies. Then we can also learn more about how medicines are affecting us.
|
|
|
|
« Last Edit: June 29, 2017, 02:11:53 pm by Zanthius »
|
Logged
|
|
|
|
Julie.chan
*Many bubbles*
Offline
Posts: 130
Sharing is good.
|
Right, your "theory". Your armchair theory, that is. Just because there are a couple more people sitting in armchairs with you doesn't mean it's any more rigorous. People have been doing research for centuries, dedicating their entire lives to it. I don't think someone like you or I is qualified to tell them how to properly do research. They know better than we do.
|
|
|
|
|
Logged
|
|
|
|
Zanthius
Enlightened
Offline
Posts: 941
|
Right, your "theory". Your armchair theory, that is. Just because there are a couple more people sitting in armchairs with you doesn't mean it's any more rigorous. People have been doing research for centuries, dedicating their entire lives to it.
Of course people have been doing research for centuries, but we haven't had Internet for centuries, and it is first now that we can collect information like this. Big data research also isn't my theory (https://www.journals.elsevier.com/big-data-research/).
Or here. "Big data meets public health", published in Science in 2014: http://science.sciencemag.org/content/346/6213/1054.long
I don't think someone like you or I is qualified to tell them how to properly do research. They know better than we do.
Ehh... You don't know anything about me. I am actually a chemistry researcher, and part of my work involves researching how a type of molecules called anthocyanins affect humans.
I don't think most researchers think this is a bad idea. I think people that are overly concerned about their privacy tend to think this is a bad idea.
Also, I don't think you should believe in authorities, unless they can explain to you why you should do it like them. Many authorities do things just because everybody else does, and they don't necessarily even know why they are doing it like that. There are tons of people with PhD's that can't think "out-of-the-box", and just do things in the traditional way because that is what they were taught. The traditional educational system encourages conformity, and discourages divergent thinking (in some cases even independent thinking).
|
|
|
|
« Last Edit: June 29, 2017, 08:20:09 pm by Zanthius »
|
Logged
|
|
|
|
|
|
Scalare
*Many bubbles*
Offline
Posts: 245
|
Zanthius, if you keep hooking up things to a neural net like you do you will essentially get a 360 degree recommendation engine ,totally tailored to your body and persona.
This will create the best way to fight cancer, generate a neural net teacher to teach you anything you want, etc etc. But it will also detect diseases for you. Any desease. It will predict your death, and that of your children, with great accuracy. You will know the future and you will know that for example your kid A has 3 years left to live, and your wife will never get past 60, while you get to live to 95. And that kid will know as well, because he has simply asked his recommender when he will die.
But what if you take this worldwide with world data aggregated from every country in the world (instead of like how it is now you only get news which has been thorougly americanized and are thusly an indoctrinated fool simply by turning on the internet, phone or tv in the USA).
You would get a universal truth engine. It can tell you very inconvenient truths, for example that the Bush administration planned 9/11, or it can tell your wife with 95% certainty that you are cheating on her. And the justice system will start using it to determine by statistics if you did a certain crime, and even predict certain crimes.
That's also the danger of hooking up all data to an algorithm. You get an universal truth engine, that doesn't know whether it reached that conclusion by statistics or by facts.
I applaud your efforts to connect the data already out there, basically AI = algorithm x computer power x data. If you increase data x2 your AI will be 2x stronger, same with an 2x more efficient algorithm or computer power being 2x more powerful.
So this will effectively make better AI a possibility.
|
|
|
|
|
Logged
|
|
|
|
Julie.chan
*Many bubbles*
Offline
Posts: 130
Sharing is good.
|
I am actually a chemistry researcher
Good for you. So you should understand that you can't assume a causation based on a correlation. A probable mechanism has to be predicted and tested.
This is hard in nutrition. Working with chemicals, all you need to do is perform experiments with chemicals. With nutrition, you can't do such a simple randomized controlled trial, because you can't possibly control everything everyone eats for the years or even decades it would take. Plus there are ethics concerns you don't need to worry about in your field.
I don't think most researchers think this is a bad idea. I think people that are overly concerned about their privacy tend to think this is a bad idea.
That's great and all, but there's a difference between not thinking that something is a bad idea and thinking it can substitute for randomized controlled trials.
As for privacy, it's not my privacy. It's our privacy. I don't want the United States to be a surveillance state. I don't want people to be unable to keep secrets, unable to not be watched, unable to not be listened to. In order to collect all this data you want to collect, we would need to establish draconian laws requiring everyone to use an ID card to purchase anything, eat anything, go anywhere, etc. We would also have to establish surveillance cameras everywhere, even in people's own homes. Sound familiar? All this for what you presume might result in an incremental improvement on our understanding of nutrition. Such a tiny benefit, or even any benefit at all, is not worth this cost.
I don't think you should believe in authorities
I don't believe in authorities either. What matters is the claim, and what evidence backs up that claim.
But saying that you have, out of all the thousands before you actually working in the field you are criticizing, found out exactly what they are doing wrong and have a simple solution, you'll have to pardon me for being skeptical. If you think you can turn around the way nutrition research is done and find all these answers we've been struggling to find for decades, go for it. Write a paper and submit it for publication, just like you would a paper on chemistry. And if you believe that correlation can substitute for randomized controlled trials, show that being done.
In other words: less talking, more doing.
|
|
|
|
« Last Edit: June 30, 2017, 03:15:20 am by JulieMaru-chan »
|
Logged
|
|
|
|
Death 999
Global Moderator
Enlightened
Offline
Gender:
Posts: 3874
We did. You did. Yes we can. No.
|
Scalare, I think the risk you describe is not exactly a proximate risk of the endeavor he's setting out on… you know?
|
|
|
|
|
Logged
|
|
|
|
Zanthius
Enlightened
Offline
Posts: 941
|
Good for you. So you should understand that you can't assume a causation based on a correlation. A probable mechanism has to be predicted and tested.
The way I have proposed to research nutrition from big data (the entire population), doesn't say anything of a possible mechanism. That is a completely different field of study. What you can find from "big data", are things that are NOT causes, but JUST correlations. If you have 50 different types of food correlating with a health condition, it would be rather cumbersome to investigate all of them for a possible mechanism. If however you find out that 45 of those correlations CANNOT be causes, since they ONLY correlate with the consumption of other types of food that correlate with the health condition, you are left with only 5 different foods. Trying to find a mechanism from those 5 different types of food, is much more easy than trying to find a mechanism from those 50 different types of food. Maybe for example there is a certain molecule that only is present in significant amounts in those 5 different types of food. But also, human biology is very complex, and thinking that you know something about if a molecule is healthy or not, just from knowing a possible mechanism, is also very dangerous. There can be cascade effects, and the "long term" effect on a biological system can be extremely difficult to predict. Also, since there are tons of different molecules in each type of food, one of them might be healthy while another one might be unhealthy. Coffee is a good example. A molecule called cafestol was found to increase serum cholesterol, but there are also lots of antioxidants in coffee, and molecules that we assume are beneficial to your health. Fortunately, most of the cafestol is removed in the filter of filtered coffee. Anyhow, it is the overall mixture of molecules in a type of food which determines if it is healthy or not.
That's great and all, but there's a difference between not thinking that something is a bad idea and thinking it can substitute for randomized controlled trials.
Do you understand why the nutritional researchers are using randomized controlled trials? It seems to me like you only "think" using a random selection of people is a "good idea", since that is what people have been doing for a long time. Or like if they somehow discovered that using a random selection of people gave the most reliable results from doing lots of experiments. People didn't discover this empirically, they understood it from advances in basic probability and statistics during the 17th and 18th century. Mathematics is not an empirical discipline.
You should read this article by Eliezer Yudkowsky: http://lesswrong.com/lw/iq/guessing_the_teachers_password/
As for privacy, it's not my privacy. It's our privacy. I don't want the United States to be a surveillance state. I don't want people to be unable to keep secrets, unable to not be watched, unable to not be listened to. In order to collect all this data you want to collect, we would need to establish draconian laws requiring everyone to use an ID card to purchase anything, eat anything, go anywhere, etc.
I don't know how things are in your country, but here we almost never use cash anymore. When I use my debit card to pay in grocery stores, I am already giving my ID to the payment terminal. The only people that still are using cash in my country, are criminals and very old people. The politicians here are discussing to get completely rid of cash, since they believe it can reduce black market activity.
We would also have to establish surveillance cameras everywhere, even in people's own homes. Sound familiar?
Ehhhh... where have I said anything about using surveillance cameras to collect data?
|
|
|
|
« Last Edit: June 30, 2017, 12:54:45 pm by Zanthius »
|
Logged
|
|
|
|
|
