I agree it wasn't ideal but that's a bigger problem if you're trying to gauge what percentage of the background population in general considers breastfeeding images to be inappropriate.
I was trying to gauge the difference in people's reactions to a black woman's photo and a white woman's photo, and so even if the wording may bias respondents one way or another, I don't think that's as big of a problem if it weighs equally on both sides.
In any case, I think the problem you describe was unavoidable in this case, because if the whole point of your survey is to ask people, "Is this image inappropriate?", then just by asking the question you have already put into their minds the notion that it might be inappropriate. (The fictional context I described was close to the real-life contexts that the real pictures were published in.) There may have been a better way of asking it though. How would you have done it?
In this case, the survey on Mechanical Turk was limited to U.S. users only. I did say that in the article:) ("...limited to U.S. users...") But you certainly make a good point generally.
Even with small samples, if you get very little difference in the percentages for your two samples, you can still conclude, very probably, that there is probably no big difference in the background populations.
For example, we know almost certainly that it's not the case that 70% of the background population would be offended by the black woman's photo but only 30% would be offended by the white woman's photo, because under that assumption, it's very unlikely we would have gotten around 70% in both samples, even with samples of only size 50.
Of course there's not enough evidence to reject the hypothesis that the background percentages might be, say, 70% and 75%.
That would be pretty cool, but I don't know if the incremental gain would be worth it.
By switching from centralized social networking (Facebook) to decentralized (Diaspora), the big change is that whereas previously your content would get removed if it displeased only one entity (Facebook), now you can host any content as long as you can find one hosting provider anywhere in the world who is willing to host it.
By switching to decentralized (Diaspora) to a decentralized, distributed model (some kind of BitTorrentFacebook), now you've added the benefit that you can also post stuff that no hosting provider anywhere in the world would be willing to host. (This is in fact what BitTorrent often gets used for, serving files that no hosting provider will host for you, either because they violate copyright laws or because they consume too much bandwidth.) Whether that's worth it or not, depends on how much value you assign to those pieces of content that "nobody would be willing to host".
But people need a way to actually access your profile. It would be hard to come up with a protocol to host your profile "on your cell phone number", as opposed to hosting it at a webpage which can be accessed under your domain.
The company that creates and launches the open, decentralized social networking protocol, could set themselves up as the default node for people to create new profiles, and could make money by running ads somewhere on those profiles. There's nothing in my suggestion for an open social networking protocol that would preclude the profile host from running ads, if that's their agreement with the users being hosted there. Or users could pay a (presumably small) amount of money to have the ads removed from their profiles.
Good point, yes I got that press release from them after I'd filed the story: http://www.prnewswire.com/news...
and here's the charter: https://ello.co/downloads/ello...
Although as someone else pointed out, it looks like under Delaware law a 2/3 supermajority could vote to amend the charter, or to be bought out by another entity that doesn't have to honor the original charter.
The point was for the reader to see if they would independently come up with the same suggestion that I did, after reading a description of how they do things now.
Even if they use tokens, the bag-pickup would still be a bottleneck if the person taking the token then has to walk to the back of the truck, grab the bag, and bring it back, which slows the line down more than anything. If they have the bag already sitting on the counter though, that will speed up the line whether the bags are already sitting on the counter or not.
You're right. "Algorithm" was probably the wrong word. But that's probably a less important point than the question of whether a simple change could save people thousands of man-hours spent waiting in the desert sun every year.
The volunteers might not scale in a linear fashion. But it would certainly be faster. In any case you could keep scaling them up until the truck becomes the bottleneck, and then decide whether or not to keep scaling up from there, taking the diminishing returns into account.
I doubt the Nevada Health Department would give wrong information just to screw with BM, but you do have a good point that it would be a good idea to get a formal statement from them, in advance, that this was OK. The guy who answered my inquiry did say "Your questions regarding ice service at Burning Man was brought to my attention as we are the health authority for the Burning Man event", so I assume I got the right people.
As I think someone else pointed out, nobody except the official ice vendor is allowed to sell ice at Burning Man, so it's not like I can just go and do it myself...
Assuming they're in it to make a profit at all, how would it not be more profitable to sell more bags of ice per minute?
And, maybe getting the idea out there will cause it to gain more traction and they'll do it. I assume that's why Burning Man implemented the shuttle service from the airport.
I thought about putting the entire idea in the summary before the link. But I also wanted to give people a chance to see if the exact same idea would occur to them, after reading the description of how it's done now.
But you're right, I didn't like how it sounded like all those one-weird-trick ads about how to pay down your mortgage with acai berries.
Yes you can always speed up the system by having multiple "cash registers". But I think the proportional gains are much bigger by simply bringing ice to the table before the order is placed.
Suppose in the existing system, each person goes to the front, places an order, the person takes their order and goes back to the truck to get the ice from a guy handing ice bags out of the truck, and then the bag goes to the customer. Two employees and you can handle about 1 customer per minute.
If you had 12 employees instead of 2, then you could handle 6 customers per minute.
On the other hand, if you had the ice bags ready on the counter before the customer placed the order, then you'd probably only need 3 employees working concurrently (one to take the cash, two to keep the ice bags coming), and each transaction would probably come in under 10 seconds so you'd still handle 6 customers per minute.
You could always do both (more cash registers, and bags of ice pre-fetched on the counter). But if you call for more cash registers, they might object that they don't have the manpower to provide that. On the other hand there's really no good reason not to have the bags of ice on the table already ready for the customer to pick them up.
That was already my starting point with the article. The article was about how people were literally saying "This bad thing happened because she took a non-zero risk", and strongly implying that this was a criticism of her, as if she had made a mistake. And my response is that that is fallacious, because it ignores the benefits. I feel like I just said this??
It depends if you are making this statement with the implication that she should not have done it.
If you are implying that she should not have done it, then I would say that's fallacious because you're looking only at the risks, not the benefits of her action.
If you are not implying that she shouldn't have done it -- if you are simply saying "If you take a risk of a bad event happening, then the probability of that bad event is non-zero", without implying any criticism of her actions -- then of course that's true, but also so obvious that why bother saying it?
In the overwhelming majority of cases where commenters are saying "If you don't want nude photos to leak, don't take any", their tone generally implies that their comments fall in the first category (implicitly criticizing her for taking the photos), and that's the fallacy I'm attacking.
Yeah, I think you're right. Perhaps the probability of a celebrity account being hacked should not be estimated based on the probability of the average account being hacked.
It might be more appropriate to say that she correctly estimated the probability to be low, because nothing like this (large-scale hacking of celebrity cloud storage accounts) had happened before, even though presumably people are trying all the time to hack (especially female) celebrities' cloud storage accounts.
What do you mean by "secured their data" in this context?
Even if the data had been stored encrypted on Apple's servers, her cloud login password would have decrypted the files whenever she logged in, so if someone stole or brute-forced her password, they would have gotten the photos anyway.
Unless you mean encrypting the files using a separate protocol, before transmitting them to her boyfriend. It's probably safe to say that it would be a losing battle to try and persuade the majority of users to do this.
Or did you mean to simply delete the photos after they'd been shared? Perhaps, but sometimes it's hard to tell if something has been deleted for good. When you delete a file from Apple's cloud storage, does it get moved temporarily to a "trash" folder where it could be recovered later?
Perhaps the option most likely to actually be adopted by users, would be for the cloud storage company to implement a snapchat-style sharing feature, where any photos uploaded to a particular folder, will be automatically deleted (forever) after a certain time period.
Well OK, instead of "victim blaming" you could call it "telling someone that the only way to reduce a risk to zero is to not take the risk at all".
However, I would argue it's making the same logical fallacy -- looking only at the risk of an action, not the benefit.
Everybody already knows that the only way to absolutely guarantee that your nude selfies don't get out, is not to take any. If they do it anyway, it's not because they don't realize this fact. It's because there are benefits to taking nude selfies, and they're weighing the benefits.
Now you could argue that they're weighing the risks and benefits incorrectly, but that would require a separate argument.
Because if you judge the probability of an event to be very low, just because that event happens, does not mean you were incorrect. I don't think the leak was a "probable negative", because large-scale leaks of cloud photo storage, are very rare. (I assume this event was the largest ever.)
I think what you're saying is true. My point is that it's a bad analogy for the nude photo hack because: (1) Large-scale hacks of cloud storage like this, are pretty rare, so the odds of her photos being hacked were much smaller than the odds of a car being stolen with the keys in the ignition. The fact that it did happen, does not mean that JLaw was wrong to estimate that the probability was very low. (2) There are benefits to sending nude selfies, whereas there's not much benefit to leaving your keys in the ignition of your Ferrari.
Slashdot Mirror
I agree it wasn't ideal but that's a bigger problem if you're trying to gauge what percentage of the background population in general considers breastfeeding images to be inappropriate.
I was trying to gauge the difference in people's reactions to a black woman's photo and a white woman's photo, and so even if the wording may bias respondents one way or another, I don't think that's as big of a problem if it weighs equally on both sides.
In any case, I think the problem you describe was unavoidable in this case, because if the whole point of your survey is to ask people, "Is this image inappropriate?", then just by asking the question you have already put into their minds the notion that it might be inappropriate. (The fictional context I described was close to the real-life contexts that the real pictures were published in.) There may have been a better way of asking it though. How would you have done it?
In this case, the survey on Mechanical Turk was limited to U.S. users only. I did say that in the article :) ("...limited to U.S. users...") But you certainly make a good point generally.
Someone beat you to it: http://developers.slashdot.org...
The idea was that it's about Internet culture (not technology-specific) and the problem of clickbait that can be easily debunked.
Even with small samples, if you get very little difference in the percentages for your two samples, you can still conclude, very probably, that there is probably no big difference in the background populations.
For example, we know almost certainly that it's not the case that 70% of the background population would be offended by the black woman's photo but only 30% would be offended by the white woman's photo, because under that assumption, it's very unlikely we would have gotten around 70% in both samples, even with samples of only size 50.
Of course there's not enough evidence to reject the hypothesis that the background percentages might be, say, 70% and 75%.
Udderly inappropriate.
That would be pretty cool, but I don't know if the incremental gain would be worth it.
By switching from centralized social networking (Facebook) to decentralized (Diaspora), the big change is that whereas previously your content would get removed if it displeased only one entity (Facebook), now you can host any content as long as you can find one hosting provider anywhere in the world who is willing to host it.
By switching to decentralized (Diaspora) to a decentralized, distributed model (some kind of BitTorrentFacebook), now you've added the benefit that you can also post stuff that no hosting provider anywhere in the world would be willing to host. (This is in fact what BitTorrent often gets used for, serving files that no hosting provider will host for you, either because they violate copyright laws or because they consume too much bandwidth.) Whether that's worth it or not, depends on how much value you assign to those pieces of content that "nobody would be willing to host".
But people need a way to actually access your profile. It would be hard to come up with a protocol to host your profile "on your cell phone number", as opposed to hosting it at a webpage which can be accessed under your domain.
The company that creates and launches the open, decentralized social networking protocol, could set themselves up as the default node for people to create new profiles, and could make money by running ads somewhere on those profiles. There's nothing in my suggestion for an open social networking protocol that would preclude the profile host from running ads, if that's their agreement with the users being hosted there. Or users could pay a (presumably small) amount of money to have the ads removed from their profiles.
Good point, yes I got that press release from them after I'd filed the story:
http://www.prnewswire.com/news...
and here's the charter:
https://ello.co/downloads/ello...
Although as someone else pointed out, it looks like under Delaware law a 2/3 supermajority could vote to amend the charter, or to be bought out by another entity that doesn't have to honor the original charter.
The point was for the reader to see if they would independently come up with the same suggestion that I did, after reading a description of how they do things now.
Even if they use tokens, the bag-pickup would still be a bottleneck if the person taking the token then has to walk to the back of the truck, grab the bag, and bring it back, which slows the line down more than anything. If they have the bag already sitting on the counter though, that will speed up the line whether the bags are already sitting on the counter or not.
You're right. "Algorithm" was probably the wrong word. But that's probably a less important point than the question of whether a simple change could save people thousands of man-hours spent waiting in the desert sun every year.
The volunteers might not scale in a linear fashion. But it would certainly be faster. In any case you could keep scaling them up until the truck becomes the bottleneck, and then decide whether or not to keep scaling up from there, taking the diminishing returns into account.
I doubt the Nevada Health Department would give wrong information just to screw with BM, but you do have a good point that it would be a good idea to get a formal statement from them, in advance, that this was OK. The guy who answered my inquiry did say "Your questions regarding ice service at Burning Man was brought to my attention as we are the health authority for the Burning Man event", so I assume I got the right people.
As I think someone else pointed out, nobody except the official ice vendor is allowed to sell ice at Burning Man, so it's not like I can just go and do it myself...
Assuming they're in it to make a profit at all, how would it not be more profitable to sell more bags of ice per minute?
And, maybe getting the idea out there will cause it to gain more traction and they'll do it. I assume that's why Burning Man implemented the shuttle service from the airport.
I thought about putting the entire idea in the summary before the link. But I also wanted to give people a chance to see if the exact same idea would occur to them, after reading the description of how it's done now.
But you're right, I didn't like how it sounded like all those one-weird-trick ads about how to pay down your mortgage with acai berries.
Yes you can always speed up the system by having multiple "cash registers". But I think the proportional gains are much bigger by simply bringing ice to the table before the order is placed.
Suppose in the existing system, each person goes to the front, places an order, the person takes their order and goes back to the truck to get the ice from a guy handing ice bags out of the truck, and then the bag goes to the customer. Two employees and you can handle about 1 customer per minute.
If you had 12 employees instead of 2, then you could handle 6 customers per minute.
On the other hand, if you had the ice bags ready on the counter before the customer placed the order, then you'd probably only need 3 employees working concurrently (one to take the cash, two to keep the ice bags coming), and each transaction would probably come in under 10 seconds so you'd still handle 6 customers per minute.
You could always do both (more cash registers, and bags of ice pre-fetched on the counter). But if you call for more cash registers, they might object that they don't have the manpower to provide that. On the other hand there's really no good reason not to have the bags of ice on the table already ready for the customer to pick them up.
That was already my starting point with the article. The article was about how people were literally saying "This bad thing happened because she took a non-zero risk", and strongly implying that this was a criticism of her, as if she had made a mistake. And my response is that that is fallacious, because it ignores the benefits. I feel like I just said this??
It depends if you are making this statement with the implication that she should not have done it.
If you are implying that she should not have done it, then I would say that's fallacious because you're looking only at the risks, not the benefits of her action.
If you are not implying that she shouldn't have done it -- if you are simply saying "If you take a risk of a bad event happening, then the probability of that bad event is non-zero", without implying any criticism of her actions -- then of course that's true, but also so obvious that why bother saying it?
In the overwhelming majority of cases where commenters are saying "If you don't want nude photos to leak, don't take any", their tone generally implies that their comments fall in the first category (implicitly criticizing her for taking the photos), and that's the fallacy I'm attacking.
Yeah, I think you're right. Perhaps the probability of a celebrity account being hacked should not be estimated based on the probability of the average account being hacked.
It might be more appropriate to say that she correctly estimated the probability to be low, because nothing like this (large-scale hacking of celebrity cloud storage accounts) had happened before, even though presumably people are trying all the time to hack (especially female) celebrities' cloud storage accounts.
What do you mean by "secured their data" in this context?
Even if the data had been stored encrypted on Apple's servers, her cloud login password would have decrypted the files whenever she logged in, so if someone stole or brute-forced her password, they would have gotten the photos anyway.
Unless you mean encrypting the files using a separate protocol, before transmitting them to her boyfriend. It's probably safe to say that it would be a losing battle to try and persuade the majority of users to do this.
Or did you mean to simply delete the photos after they'd been shared? Perhaps, but sometimes it's hard to tell if something has been deleted for good. When you delete a file from Apple's cloud storage, does it get moved temporarily to a "trash" folder where it could be recovered later?
Perhaps the option most likely to actually be adopted by users, would be for the cloud storage company to implement a snapchat-style sharing feature, where any photos uploaded to a particular folder, will be automatically deleted (forever) after a certain time period.
Well OK, instead of "victim blaming" you could call it "telling someone that the only way to reduce a risk to zero is to not take the risk at all".
However, I would argue it's making the same logical fallacy -- looking only at the risk of an action, not the benefit.
Everybody already knows that the only way to absolutely guarantee that your nude selfies don't get out, is not to take any. If they do it anyway, it's not because they don't realize this fact. It's because there are benefits to taking nude selfies, and they're weighing the benefits.
Now you could argue that they're weighing the risks and benefits incorrectly, but that would require a separate argument.
Because if you judge the probability of an event to be very low, just because that event happens, does not mean you were incorrect. I don't think the leak was a "probable negative", because large-scale leaks of cloud photo storage, are very rare. (I assume this event was the largest ever.)
I think what you're saying is true. My point is that it's a bad analogy for the nude photo hack because: (1) Large-scale hacks of cloud storage like this, are pretty rare, so the odds of her photos being hacked were much smaller than the odds of a car being stolen with the keys in the ignition. The fact that it did happen, does not mean that JLaw was wrong to estimate that the probability was very low. (2) There are benefits to sending nude selfies, whereas there's not much benefit to leaving your keys in the ignition of your Ferrari.
to make it clear that I was quoting the mindset of the victim-blamers, and not describing what I think.