Quote websites show weakness of Google algo

[ergophobe]

<rant>No, it's not a research study. It's just a rant.

If you search for a quote, you get a huge number of hits on page after page of websites that list quotes and attribute them to someone without citation. Most quotes attributed to Twain, Churchill, Einstein, WC Fields, Oscar Wilde are apocryphal.

There are a couple of websites that actually do the work and nail down the source: Wikiquote and Quote Investigator. By any measure other than a popularity contest based on backlinks, these sites should *always* outrank Brainy Quote and Good Reads and all the rest. I have yet to see a single case where they do.

For all of Google's PR about trying to surface quality, what they really mean is that the basic, original Page Rank algorithm still mostly rules, but they do try to get rid of the worst spam. But there certainly is not any algorithm otherwise floating quality to the top. I know you all know this... just ranting, because one of the most consistently egregious cases is trying to investigate any well-known quote.

Today I was trying to determine whether the quote "I spent half my money on whiskey and women. The other half I wasted," really was a WC Fields quote. According to page after page of Google results, it is. Wikiquote for WC fields doesn't mention it . In the Talk, it is among the unsourced quotes removed from the main page.

As near as I can tell, it's actually from the film 100 Rifles, spoken by Yaqui Joe played by Burt Reynolds in 1969. There are no references in Google Books that pre-date that.

But we all know that nobody wants to  be so low-brow as to quote Burt Reynolds. So it morphs into a WC Fields quote and Google reinforces that by floating those pages to the top

</rant>

[buckworks]

Misattribution of quotes is a problem that predates the internet.

There's a quote popularly attributed to John Lennon, something like "Life is what happens while you're making other plans."

But years ago I came across that line in a book of quotes aimed at after-dinner speakers that was published in the 1950s (sorry can't remember title etc. but it was a book I borrowed from the town library in Souris, Manitoba). That book listed the quote as source unknown.

So Lennon said it but it didn't originate with him.

[ergophobe]

True, there are plenty of misattributions in the Oxford Book of Quotations and Bartlett's, but on Google misattributions completely drown out the correct attributions and it has gotten worse over time (subjective opinion).

To me it's a case study in the general problem of Google's inability to distinguish crap from quality, the fundamental poverty of the algorithm and, of course, there are other places (for a while obscure place name searches were dominated by spam, but they have actually made progress there). By contrast, while there *are* errors in the Oxford Book of Quotations, but the editors do actually know how to differentiate between good sources and bad, they just make mistakes occasionally.

[rcjordan]

Quotes, recipes, & movie/tv dialogues are perfect for easy, free, scrape-able, content spamming.  Toolman settled on building recipe sites as gateways to his pharm business.  To hell with being actually *edible* --he just needed something that looked like a recipe.

[Brad]

Google cannot distinguish quality only popularity.  Frankly, no search engine can tell quality unless there are human editors.

[rcjordan]

And when you can legally skirt copyright under 'fair use' then content spamming is at its easiest.  (Lyric sites are pushing the fair use envelope, imo.)

[ergophobe]

Yes, recipes are another good example. Though even there, there seems to be less blatant copying of error after error.

Song lyrics are another, but generally speaking, those are easier to verify since at leas they always give the artist. But yes, it's a place where widespread copying and duplicating results in a complete drowning of anything resembling quality.

>>no search engine can tell quality unless there are human editors.

Indeed. I guess my point was that for all the Google comments about AI and Rank Brain and all that, there are certain types of sites (quotes, lyrics, recipes) that lay bald how simple and bad the algorithms are. Yes, as algorithms from the point of code, they are too complex for any one person to understand. And yet from the point of the actual results, they are still sooooo much worse than even a relatively dumb human.

[rcjordan]

>less blatant copying of error after error.

You must have missed Toolman's work, then.  He'd email me cackling about the ingredients he threw in on some of them.  But his site design looked more than legit.

[ergophobe]

Maybe so. I rarely cook from recipes.

Seems like a great spammer playground though.