Putting aside Hank Green's obvious hysterics, I'm glad this author brought some rationality to the attempted outrage baiting.
But just to give a general statement, I think Google's callout snippet feature is dangerous. It's probably really fucking clever by algorithm standards and works some high percentage of the time, but that makes it even more dangerous when people have confidence that it is 100% real. I was helping somebody Google a technical problem recently, and the callout snippet suggested something that would damage the project, but this non-expert thought that it was the de-facto answer to her question because it was at the top of the search results. I think Google's design needs to make it more clear that these are algorithmically selected snippets that are not guaranteed to be accurate or solve any given question you might have.
Wait, Hank Green posted two pictures and said it was, quote, "real weird".
Which it is. It's weird. It might have a very rational explanation, but that doesn't make it not weird.
Standing outside during a solar eclipse is real weird. It's a totally explainable phenomena and once you understand it, you can rationalize the experience fairly easily. But it's still real weird.
Calling out an unusual phenomena as real weird doesn't seem to be "obvious hysterics". It's someone observing a phenomena and saying, "that's not what I expected!"
1) You're right that "obvious hysterics" isn't the best possible label, but I couldn't think of a better one.
2) I cannot read anybody's mind, but anybody who has spent any length of time online probably fully well knows that you're teeing up the Twitter outrage mob by posting that kind of comparison in the way that Hank Green did.
3) I'm not familiar with Hank Green, but I thought that the Timar Ivo Batis' response posted in that article gave a pretty good response pointing out that he knew tech better than that.
> Sorry Hank but you know better about how tech works than to draw comparisons like that. Changing wording gives you the top result for this specific wording. You really do not want a machine guessing and always giving you the same results. Too much nuance.
4) I don't want to detract further from the overall point of the article. The main takeaway from this should be that even the best algorithms can have loads of problems with figuring out true knowledge and answering questions algorithmically, and there's probably issues with Google presenting their callout snippets as the answers to questions without better disclaimers or educating people that algorithms are just guesses that can be wrong. Every establishment figure talks all of the time about the dangers of misinformation, but Google has a system that is guaranteed to present lots of information wrong. This seems worthy of critical discussion.
when used in that kind of context, it's not unreasonable for 'people' to mean 'civilization', and the arrival of western civilization seems like a more relevant event than the incas 'coming to america' if you can even say they did that
It bothers me that the answer to "When did people come to America", when clarified as something like "When did European colonists come to what is now the United States" is still wrong.
Google's answer is "The first colony was founded in Jamestown, Virginia in 1607."
But that page says: "The Spanish were among the first Europeans to explore the New World and the first to settle in what is now the United States" then gives the Jamestown example after switching focus to England and the Atlantic coast.
San Juan, PR was founded in 1521 by Spanish colonists.
St. Augustine, FL was established in 1561 by Spanish colonists and is the oldest continuously occupied settlement of European origin in the contiguous United States.
I grew up in Florida. I visited St. Augustine as a kid. We all learned that it was colonized before Jamestown. My American identity does not require English colonization be placed on an artificial pedestal.
The essay ends "after all it learned from us."
But isn't that why people were "calling Google’s search algorithm biased and racist"? Because it learned it from us?
The question that is correctly being answered is more like "when did the colonists that formed the united states first arrive".
Florida as a colony was founded by Spain, was briefly under British control and against secession, and then went back to Spain for another fifty years. Then it was finally sold to the fifty-year-old United States, much too late to be part of the founding.
8 comments
[ 2.6 ms ] story [ 27.7 ms ] threadBut just to give a general statement, I think Google's callout snippet feature is dangerous. It's probably really fucking clever by algorithm standards and works some high percentage of the time, but that makes it even more dangerous when people have confidence that it is 100% real. I was helping somebody Google a technical problem recently, and the callout snippet suggested something that would damage the project, but this non-expert thought that it was the de-facto answer to her question because it was at the top of the search results. I think Google's design needs to make it more clear that these are algorithmically selected snippets that are not guaranteed to be accurate or solve any given question you might have.
Which it is. It's weird. It might have a very rational explanation, but that doesn't make it not weird.
Standing outside during a solar eclipse is real weird. It's a totally explainable phenomena and once you understand it, you can rationalize the experience fairly easily. But it's still real weird.
Calling out an unusual phenomena as real weird doesn't seem to be "obvious hysterics". It's someone observing a phenomena and saying, "that's not what I expected!"
2) I cannot read anybody's mind, but anybody who has spent any length of time online probably fully well knows that you're teeing up the Twitter outrage mob by posting that kind of comparison in the way that Hank Green did.
3) I'm not familiar with Hank Green, but I thought that the Timar Ivo Batis' response posted in that article gave a pretty good response pointing out that he knew tech better than that.
> Sorry Hank but you know better about how tech works than to draw comparisons like that. Changing wording gives you the top result for this specific wording. You really do not want a machine guessing and always giving you the same results. Too much nuance.
4) I don't want to detract further from the overall point of the article. The main takeaway from this should be that even the best algorithms can have loads of problems with figuring out true knowledge and answering questions algorithmically, and there's probably issues with Google presenting their callout snippets as the answers to questions without better disclaimers or educating people that algorithms are just guesses that can be wrong. Every establishment figure talks all of the time about the dangers of misinformation, but Google has a system that is guaranteed to present lots of information wrong. This seems worthy of critical discussion.
Google's answer is "The first colony was founded in Jamestown, Virginia in 1607."
It links to https://www.americaslibrary.gov/jb/colonial/jb_colonial_subj... .
But that page says: "The Spanish were among the first Europeans to explore the New World and the first to settle in what is now the United States" then gives the Jamestown example after switching focus to England and the Atlantic coast.
San Juan, PR was founded in 1521 by Spanish colonists.
St. Augustine, FL was established in 1561 by Spanish colonists and is the oldest continuously occupied settlement of European origin in the contiguous United States.
I grew up in Florida. I visited St. Augustine as a kid. We all learned that it was colonized before Jamestown. My American identity does not require English colonization be placed on an artificial pedestal.
The essay ends "after all it learned from us."
But isn't that why people were "calling Google’s search algorithm biased and racist"? Because it learned it from us?
Florida as a colony was founded by Spain, was briefly under British control and against secession, and then went back to Spain for another fifty years. Then it was finally sold to the fifty-year-old United States, much too late to be part of the founding.