<?xml version="1.0" encoding="UTF-8"?><rss version="2.0"
	xmlns:content="http://purl.org/rss/1.0/modules/content/"
	xmlns:dc="http://purl.org/dc/elements/1.1/"
	xmlns:atom="http://www.w3.org/2005/Atom"
	xmlns:sy="http://purl.org/rss/1.0/modules/syndication/"
	
	>
<channel>
	<title>Comments on: Lies, Damned Lies, And Facebook (Part 2 of ∞)</title>
	<atom:link href="http://slatestarcodex.com/2013/04/04/lies-damned-lies-and-facebook-part-2-of-%E2%88%9E/feed/" rel="self" type="application/rss+xml" />
	<link>http://slatestarcodex.com/2013/04/04/lies-damned-lies-and-facebook-part-2-of-%e2%88%9e/</link>
	<description>In a mad world, all blogging is psychiatry blogging</description>
	<lastBuildDate>Fri, 24 Jul 2015 10:12:43 +0000</lastBuildDate>
	<sy:updatePeriod>hourly</sy:updatePeriod>
	<sy:updateFrequency>1</sy:updateFrequency>
	<generator>http://wordpress.org/?v=4.2.3</generator>
	<item>
		<title>By: Avery</title>
		<link>http://slatestarcodex.com/2013/04/04/lies-damned-lies-and-facebook-part-2-of-%e2%88%9e/#comment-90066</link>
		<dc:creator><![CDATA[Avery]]></dc:creator>
		<pubDate>Thu, 29 May 2014 06:05:34 +0000</pubDate>
		<guid isPermaLink="false">http://slatestarcodex.com/?p=351#comment-90066</guid>
		<description><![CDATA[Thank you, good sir, for inspiring the best laugh I&#039;ve had in a good long while. I just about lost it at &quot;Dead people should stay dead,&quot; interfering greatly with my attempt to read the whole thing aloud to my husband.]]></description>
		<content:encoded><![CDATA[<p>Thank you, good sir, for inspiring the best laugh I&#8217;ve had in a good long while. I just about lost it at &#8220;Dead people should stay dead,&#8221; interfering greatly with my attempt to read the whole thing aloud to my husband.</p>
<p><a href="javascript:void(0)" onclick="report_comments_flag(this, '90066', '4b33b77030')" class="report-comment">Report comment</a></p>
]]></content:encoded>
	</item>
	<item>
		<title>By: honnibal</title>
		<link>http://slatestarcodex.com/2013/04/04/lies-damned-lies-and-facebook-part-2-of-%e2%88%9e/#comment-20914</link>
		<dc:creator><![CDATA[honnibal]]></dc:creator>
		<pubDate>Sun, 17 Nov 2013 03:54:23 +0000</pubDate>
		<guid isPermaLink="false">http://slatestarcodex.com/?p=351#comment-20914</guid>
		<description><![CDATA[I doubt anybody will be able to comment publicly on exactly how Google&#039;s search completion works. But, here&#039;s a plausible scenario.

Check out the results here: https://encrypted.google.com/#q=french+people+should+die

See how the second link is &quot;Judge Agrees Marine Should Die For Murdering French Valley Couple&quot;? It&#039;s a useful reminder that Google will be chopping up the input query and searching for its component words and phrases individually, as well as the whole thing.

A natural way to model the probability of the next word after &quot;French people should...&quot; is to make a maximum likelihood estimate: given some example data, count the probability of each continuation, and normalise by the probability of the original.

But language is pretty sparse, so no sample is ever big enough, once you&#039;re asked to model continuations from more than a few words. So you get better results by asking questions about the distribution of words immediately following &quot;should&quot;, searching for &quot;french people&quot;, etc.

So you&#039;ll see similar results for &quot;Republicans should...&quot; and &quot;Democrats should...&quot;, because both results are strongly influenced by the query &quot;should...&quot;.]]></description>
		<content:encoded><![CDATA[<p>I doubt anybody will be able to comment publicly on exactly how Google&#8217;s search completion works. But, here&#8217;s a plausible scenario.</p>
<p>Check out the results here: <a href="https://encrypted.google.com/#q=french+people+should+die" rel="nofollow">https://encrypted.google.com/#q=french+people+should+die</a></p>
<p>See how the second link is &#8220;Judge Agrees Marine Should Die For Murdering French Valley Couple&#8221;? It&#8217;s a useful reminder that Google will be chopping up the input query and searching for its component words and phrases individually, as well as the whole thing.</p>
<p>A natural way to model the probability of the next word after &#8220;French people should&#8230;&#8221; is to make a maximum likelihood estimate: given some example data, count the probability of each continuation, and normalise by the probability of the original.</p>
<p>But language is pretty sparse, so no sample is ever big enough, once you&#8217;re asked to model continuations from more than a few words. So you get better results by asking questions about the distribution of words immediately following &#8220;should&#8221;, searching for &#8220;french people&#8221;, etc.</p>
<p>So you&#8217;ll see similar results for &#8220;Republicans should&#8230;&#8221; and &#8220;Democrats should&#8230;&#8221;, because both results are strongly influenced by the query &#8220;should&#8230;&#8221;.</p>
<p><a href="javascript:void(0)" onclick="report_comments_flag(this, '20914', '4b33b77030')" class="report-comment">Report comment</a></p>
]]></content:encoded>
	</item>
	<item>
		<title>By: DSimon</title>
		<link>http://slatestarcodex.com/2013/04/04/lies-damned-lies-and-facebook-part-2-of-%e2%88%9e/#comment-17614</link>
		<dc:creator><![CDATA[DSimon]]></dc:creator>
		<pubDate>Thu, 17 Oct 2013 16:29:45 +0000</pubDate>
		<guid isPermaLink="false">http://slatestarcodex.com/?p=351#comment-17614</guid>
		<description><![CDATA[You may also need to delete all your Google cookies.]]></description>
		<content:encoded><![CDATA[<p>You may also need to delete all your Google cookies.</p>
<p><a href="javascript:void(0)" onclick="report_comments_flag(this, '17614', '4b33b77030')" class="report-comment">Report comment</a></p>
]]></content:encoded>
	</item>
	<item>
		<title>By: Anonymous</title>
		<link>http://slatestarcodex.com/2013/04/04/lies-damned-lies-and-facebook-part-2-of-%e2%88%9e/#comment-15963</link>
		<dc:creator><![CDATA[Anonymous]]></dc:creator>
		<pubDate>Sat, 17 Aug 2013 21:26:30 +0000</pubDate>
		<guid isPermaLink="false">http://slatestarcodex.com/?p=351#comment-15963</guid>
		<description><![CDATA[*applauds*]]></description>
		<content:encoded><![CDATA[<p>*applauds*</p>
<p><a href="javascript:void(0)" onclick="report_comments_flag(this, '15963', '4b33b77030')" class="report-comment">Report comment</a></p>
]]></content:encoded>
	</item>
	<item>
		<title>By: Oligopsony</title>
		<link>http://slatestarcodex.com/2013/04/04/lies-damned-lies-and-facebook-part-2-of-%e2%88%9e/#comment-2435</link>
		<dc:creator><![CDATA[Oligopsony]]></dc:creator>
		<pubDate>Fri, 05 Apr 2013 23:42:36 +0000</pubDate>
		<guid isPermaLink="false">http://slatestarcodex.com/?p=351#comment-2435</guid>
		<description><![CDATA[whatdoestheinternetthink.net may serve as a control for Google (unless based on it in some way; I don&#039;t know.)]]></description>
		<content:encoded><![CDATA[<p>whatdoestheinternetthink.net may serve as a control for Google (unless based on it in some way; I don&#8217;t know.)</p>
<p><a href="javascript:void(0)" onclick="report_comments_flag(this, '2435', '4b33b77030')" class="report-comment">Report comment</a></p>
]]></content:encoded>
	</item>
	<item>
		<title>By: Mary</title>
		<link>http://slatestarcodex.com/2013/04/04/lies-damned-lies-and-facebook-part-2-of-%e2%88%9e/#comment-2407</link>
		<dc:creator><![CDATA[Mary]]></dc:creator>
		<pubDate>Fri, 05 Apr 2013 19:19:32 +0000</pubDate>
		<guid isPermaLink="false">http://slatestarcodex.com/?p=351#comment-2407</guid>
		<description><![CDATA[Probably revolving about the whole Indo-European stuff.]]></description>
		<content:encoded><![CDATA[<p>Probably revolving about the whole Indo-European stuff.</p>
<p><a href="javascript:void(0)" onclick="report_comments_flag(this, '2407', '4b33b77030')" class="report-comment">Report comment</a></p>
]]></content:encoded>
	</item>
	<item>
		<title>By: Avantika</title>
		<link>http://slatestarcodex.com/2013/04/04/lies-damned-lies-and-facebook-part-2-of-%e2%88%9e/#comment-2361</link>
		<dc:creator><![CDATA[Avantika]]></dc:creator>
		<pubDate>Fri, 05 Apr 2013 08:07:32 +0000</pubDate>
		<guid isPermaLink="false">http://slatestarcodex.com/?p=351#comment-2361</guid>
		<description><![CDATA[I&#039;m getting &#039;Indians are caucasians&#039; which seems utterly bizarre.]]></description>
		<content:encoded><![CDATA[<p>I&#8217;m getting &#8216;Indians are caucasians&#8217; which seems utterly bizarre.</p>
<p><a href="javascript:void(0)" onclick="report_comments_flag(this, '2361', '4b33b77030')" class="report-comment">Report comment</a></p>
]]></content:encoded>
	</item>
	<item>
		<title>By: Brian Delwiche</title>
		<link>http://slatestarcodex.com/2013/04/04/lies-damned-lies-and-facebook-part-2-of-%e2%88%9e/#comment-2359</link>
		<dc:creator><![CDATA[Brian Delwiche]]></dc:creator>
		<pubDate>Fri, 05 Apr 2013 07:36:12 +0000</pubDate>
		<guid isPermaLink="false">http://slatestarcodex.com/?p=351#comment-2359</guid>
		<description><![CDATA[I&#039;m getting &quot;Sauron shoulda put a ring on it&quot;, which I think I like better.]]></description>
		<content:encoded><![CDATA[<p>I&#8217;m getting &#8220;Sauron shoulda put a ring on it&#8221;, which I think I like better.</p>
<p><a href="javascript:void(0)" onclick="report_comments_flag(this, '2359', '4b33b77030')" class="report-comment">Report comment</a></p>
]]></content:encoded>
	</item>
	<item>
		<title>By: Julia</title>
		<link>http://slatestarcodex.com/2013/04/04/lies-damned-lies-and-facebook-part-2-of-%e2%88%9e/#comment-2354</link>
		<dc:creator><![CDATA[Julia]]></dc:creator>
		<pubDate>Fri, 05 Apr 2013 01:41:57 +0000</pubDate>
		<guid isPermaLink="false">http://slatestarcodex.com/?p=351#comment-2354</guid>
		<description><![CDATA[Jeff did an elaborate test of variations on &quot;Why are [nationality] so ....&quot;, which comes up with uniformly negative questions.  First hit for Americans is &quot;Why are Americans so stupid?&quot; followed by &quot;so cocky,&quot; &quot;so lazy,&quot; and &quot;so rude.&quot;

Canadians seem to have the best Google reputation, as people want to know why they are &quot;so polite&quot; and &quot;so friendly&quot;.
http://www.jefftk.com/news/2012-08-17]]></description>
		<content:encoded><![CDATA[<p>Jeff did an elaborate test of variations on &#8220;Why are [nationality] so &#8230;.&#8221;, which comes up with uniformly negative questions.  First hit for Americans is &#8220;Why are Americans so stupid?&#8221; followed by &#8220;so cocky,&#8221; &#8220;so lazy,&#8221; and &#8220;so rude.&#8221;</p>
<p>Canadians seem to have the best Google reputation, as people want to know why they are &#8220;so polite&#8221; and &#8220;so friendly&#8221;.<br />
<a href="http://www.jefftk.com/news/2012-08-17" rel="nofollow">http://www.jefftk.com/news/2012-08-17</a></p>
<p><a href="javascript:void(0)" onclick="report_comments_flag(this, '2354', '4b33b77030')" class="report-comment">Report comment</a></p>
]]></content:encoded>
	</item>
	<item>
		<title>By: BeoShaffer</title>
		<link>http://slatestarcodex.com/2013/04/04/lies-damned-lies-and-facebook-part-2-of-%e2%88%9e/#comment-2351</link>
		<dc:creator><![CDATA[BeoShaffer]]></dc:creator>
		<pubDate>Thu, 04 Apr 2013 22:46:26 +0000</pubDate>
		<guid isPermaLink="false">http://slatestarcodex.com/?p=351#comment-2351</guid>
		<description><![CDATA[I get &quot;rationalists should win&quot;.]]></description>
		<content:encoded><![CDATA[<p>I get &#8220;rationalists should win&#8221;.</p>
<p><a href="javascript:void(0)" onclick="report_comments_flag(this, '2351', '4b33b77030')" class="report-comment">Report comment</a></p>
]]></content:encoded>
	</item>
</channel>
</rss>
