Earlier quoted context omitted.
On Google Now, "news" sites like infowars, globalresearch.ca and on reign supreme, unless you manually filter them out. This hoax about "staged" MH17 crash is on my newsfeed why.. oh right "because you displayed interest in Ukrainian conflict". Fair enough then! It is scary how much the default app on Android serves to legitimize the fringe.
Welcome to your very own filter bubble.
Google featured snippets are worse than fake news
61–70 of 192 posts
Re: Google featured snippets are worse than fake news
#62Why is it that hard for engineers to rely on good old attribution? If every Google's featured snippet started its reply with "Breitbart says..." or "Trent Online, the leading Internet Newspaper in Nigeria said...", it wouldn't matter so much for those inevitable cases when the reply is taken straight from a white-supremacist or radical anarchist forum. The problem comes when the same reply is provided as "Google's tr…
Re: Google featured snippets are worse than fake news
#63Earlier quoted context omitted.
Problem is, can you algorithmically determine that breitbart.com and the likes can't be trusted , if they get tons of links, mentions, etc. just like the reputable news sources?
Maybe all linkers shouldn't be treated as equal. Random FB/Twitter posts reveal interest in the topic, but not accuracy of the article. But I assumed Google was accounting for this already.
However...
1. Various 'credible' news sources also link to dodgy sites and articles, either when criticising it, posting a story said sites first ran or simply because they've fallen for 'fake news'.
2. A lot of trust signals are easily gamed, like .edu links going for a fortune on SEO forums.
3. At some point, quantity will probably win out over quality.
Re: Google featured snippets are worse than fake news
#64And then there's the problem that CNN would be "fake news" because you can't exactly provide a citation for "anonymous high ranking intelligence community officials." But then... maybe that's a good thing.
Re: Google featured snippets are worse than fake news
#65Earlier quoted context omitted.
> will hinder their ability to be a source for gaining knowledge CAUTION: subjective experience ahead I feel like in the last few years Google's utility for knowledge discovery has decreased. Instead it seems to be geard more and more to drive you towards purchase and shallow content sites. I think the growing popularity of "awesome" lists and other forms of curated content discovery is also driven in part by Google'…
Google stopped providing relevant search results more than 10 years ago. It used to be several pages of relevant search results, now you are lucky if you have more than 2 or 3 somewhat related results in the first page and that's about it. You can tell there was a paradigm shift at some point and the search engine became a second (third? fourth?) class citizen.
Re: Google featured snippets are worse than fake news
#66Earlier quoted context omitted.
Google stopped providing relevant search results more than 10 years ago. It used to be several pages of relevant search results, now you are lucky if you have more than 2 or 3 somewhat related results in the first page and that's about it. You can tell there was a paradigm shift at some point and the search engine became a second (third? fourth?) class citizen.
Actually google has become ( for me, at least) lately some kind of proxy to stackoverflow and wikipedia, since both this two services have atrocious own search solutions and I have to rely on google to get interpret my keywords the right way.
Re: Google featured snippets are worse than fake news
#67Calorie counts are another that google often gets wrong. Try "Calories in corn" or "calories in black beans."
Re: Google featured snippets are worse than fake news
#68Google has developed advanced NLP algorithms, so it should be easy for them to corroborate such data before it gets pushed to featured snippets. On second thought, maybe they're already doing it and sites including authority are just copying off each other.
What makes you think that "advanced NLP algorithms" can distinguish true statements from false ones? They're not even close. You'd have to dramatically simplify the scope of the problem to make it not require artificial general intelligence. The "Fake News Challenge" [1] is the first attempt I know of to seriously evaluate NLP that could eventually be able to fact-check real-world statements, and the current task is…
Re: Google featured snippets are worse than fake news
#69The idea takes hold in the collective Google-consciousness that a fact is whatever the majority of their users believe to be a fact. There is a precedent for this, after all. They defined Spam as whatever their users said was Spam.
At first there is some friction between the knowledge of internal Google personnel (especially down in that pesky engineering department) and the new shared reality developing out in Userland. However, once an appropriate terminology is developed (in-facts and ex-facts) even the engineers are satisfied - after all their main concern was being able to draw nice clear lines between things.
Meanwhile in Userland while the boiling point of water is largely unchanged, the death toll in the Biafran blockade has shrunk to approximately 17 persons (with some arguments over whether to count people who were over 80 when it started).
Then one day an Engineer has a neatO idea which will eliminate the need for storing two sets of facts and thus save valuable Petabytes of storage and, more importantly, significant code complexity. He calls this idea Authority. Pitching it to his superiors in the marketing department he explains: "You see opinions are like assholes" - (there's a collective wrinkling of marketing noses) - "everybody's got one!". He goes on to explain that some people have more knowledge and are more scrupulous about accepting new facts than other people. Authority would identify such users and give greater weight to their activity and feedback. A profound silence falls over the room. A voice comes over the meeting room Intercom:
"Are you saying...are you suggesting...that some of our users are better than others?"
"Not better, " says the cowed Engineer. "Just more...um...authoritative. Purely in an informational sense."
"Authoritative." The voice draws the word out as if profoundly contemplating its meaning. "OK. We're going to submit this idea to our Hard AI (Larrey) - the secret one we're afraid to network."
Everyone waits. After a minute the voice says:
"Larrey has a question for you, Engineer. It is this: If you're so smart, how come you ain't rich?"
Re: Google featured snippets are worse than fake news
#70Why is it that hard for engineers to rely on good old attribution? If every Google's featured snippet started its reply with "Breitbart says..." or "Trent Online, the leading Internet Newspaper in Nigeria said...", it wouldn't matter so much for those inevitable cases when the reply is taken straight from a white-supremacist or radical anarchist forum. The problem comes when the same reply is provided as "Google's tr…
Because prepending that doesn't solve the problem Google wants to solve: once Google figures out that you typed in a question, they want to give you the answer to your question. Not tell you what other people think the answer is. (I agree that it is broken in a lot of cases now, but prepending that is a very bad workaround).