Re: Truthiness

1

I'm not super confident that their algorithm is really that good, but anything that gives climate change deniers a conniption is OK with me.

I also love how the right wing media watchdog is all upset because this will tend to favor liberal ideas over conservative ones. It's that damn reality-based thing again.


Posted by: togolosh | Link to this comment | 07-22-15 7:45 AM
horizontal rule
2

Seems gameable.


Posted by: Sifu Tweety | Link to this comment | 07-22-15 7:47 AM
horizontal rule
3

Everything is gameable. But it seems reasonable that this would inch things in the right direction.


Posted by: heebie-geebie | Link to this comment | 07-22-15 7:50 AM
horizontal rule
4

All search ranking is at some level gameable, but Google spends a lot of effort counter-gaming, and I think we generally stay ahead.


Posted by: Nathan Williams | Link to this comment | 07-22-15 8:12 AM
horizontal rule
5

I dispute the veracity of the statement that everything is gameable.


Posted by: SP | Link to this comment | 07-22-15 8:15 AM
horizontal rule
6

SP has been hacked!


Posted by: heebie-geebie | Link to this comment | 07-22-15 8:24 AM
horizontal rule
7

Wasn't there a post on this at the time?


Posted by: Ginger Yellow | Link to this comment | 07-22-15 8:32 AM
horizontal rule
8

You're right, someone stole my highly-classified unfogged.com login info.


Posted by: SP | Link to this comment | 07-22-15 8:35 AM
horizontal rule
9

7: You mean earlier this morning?


Posted by: heebie-geebie | Link to this comment | 07-22-15 8:37 AM
horizontal rule
10

This sounds like another one of those things where Google announces something that sounds potentially very creepy and totalitarian and where their main justification for the claim that it won't turn out that way is that, hey, they're Google and they have a "don't be evil" slogan!

That said I really, really dislike the various anti-science movements and I'd be willing to sacrifice stuff in order to make their recruitment efforts substantially less effective and their lives worse. So I'm ambivalent here.

Using consensus as a way to measure accuracy sounds exactly like the sort of thing you get when a bunch of people try to find a proxy they can measure more easily in order to get at something which is very hard to measure, and get caught up in groupthink about it rather than realizing that things that are hard to measure are often hard to measure because they're hard to measure and easy to measure things are going to be very different as a result.


Posted by: MHPH | Link to this comment | 07-22-15 8:38 AM
horizontal rule
11

I'm not super-skeptical about this, but I do share the concerns of 10.3. The term bien-pensant seems relevant here.


Posted by: JRoth | Link to this comment | 07-22-15 8:45 AM
horizontal rule
12

like the sort of thing you get when a bunch of people try to find a proxy they can measure more easily in order to get at something which is very hard to measure, and get caught up in groupthink about it rather than realizing that things that are hard to measure are often hard to measure because they're hard to measure and easy to measure things are going to be very different as a result.

Oh god, faculty meetings and accountability conversations are just around the corner, aren't they.


Posted by: heebie-geebie | Link to this comment | 07-22-15 8:46 AM
horizontal rule
13

7: You mean earlier this morning?

No, I mean back in March.


Posted by: Ginger Yellow | Link to this comment | 07-22-15 9:13 AM
horizontal rule
14

Using consensus as a way to measure accuracy? Consensus among whom? Generally it's consensus among people willing to waste enough time endlessly arguing. This is good for opposing anti-science movements, but only because libertarian Wikipedia editor types are opposed to anti-science movements.


Posted by: Cryptic ned | Link to this comment | 07-22-15 9:18 AM
horizontal rule
15

3, 4: yeah, I mean, having thought about it more, insofar as google's remit is "return useful results from search" keeping "measles" from being overwhelmed by "VACCINES CAUSE DROPSY" seems like a reasonable-enough tweak. Just don't hire Dr. Oz to vet.


Posted by: Sifu Tweety | Link to this comment | 07-22-15 9:21 AM
horizontal rule
16

All search ranking is at some level gameable, but Google spends a lot of effort counter-gaming, and I think we generally stay ahead.

My sense of Google's changes to its search algorithm and decisions about what to favor and what to disfavor don't make me all that sanguine about this idea.


Posted by: nosflow | Link to this comment | 07-22-15 9:26 AM
horizontal rule
17

"Dropsy" sounds so cute. Aw, who's got a little dropsy-wopsy!


Posted by: nosflow | Link to this comment | 07-22-15 9:28 AM
horizontal rule
18

There should really be a line of teddy bears for the great historical communicable diseases, and a cartoon tie-in.


Posted by: conflated | Link to this comment | 07-22-15 10:09 PM
horizontal rule
19

18: Well, there's this...


Posted by: teofilo | Link to this comment | 07-22-15 10:14 PM
horizontal rule
20

So two parallel sets of characters, one macro, one micro, around the same plot? Like Care Bears meets epidemiological Herman's Head.


Posted by: conflated | Link to this comment | 07-22-15 11:12 PM
horizontal rule
21

It is a very slippery and dangerous slope because there's no arguing with a machine

I'd pay cash money to watch that halfbright small-town weatherboy prick futilely pleading with the plummeting curve on his Google Analytics account.


Posted by: Alex | Link to this comment | 07-23-15 6:24 AM
horizontal rule
22

yeah, I mean, having thought about it more, insofar as google's remit is "return useful results from search" keeping "measles" from being overwhelmed by "VACCINES CAUSE DROPSY" seems like a reasonable-enough tweak

This is actually an interesting point. If you treat search as an application that you use to find web pages matching some arbitrary-but-structured string - like everyone who developed the skill of creating good search strings and interpreting the results from AltaVista to about 2008, i.e. Unfogged readers - you'll want just the raw results, thanks.

But if you treat it as a magic box that you type (or yell OK GOOGLE!) FACEBOOK in for Facebook and MEASLES if you think you've got measles and you always hit the first result - which everyone else does, and which Google has engineered for since Caffeine at least, well, there's a much stronger case for moderation.

I for one would love to subscribe to "Google Classic", pre-Caffeine, pre-Instant, pre-Demand Media-induced tweaks.


Posted by: Alex | Link to this comment | 07-23-15 6:30 AM
horizontal rule
23

Google has really never worked that way, though. From the beginning PageRank was designed to do more filtering (where "filtering" mostly gets expressed as "ordering of results") than that. That's how they killed AltaVista et al -- finding web pages that match an arbitrary but structured string had pretty much stopped working effectively on its own (without some intelligence in ordering) by 2005. The fact that google since because so much of a market mover that people started trying to build specifically PageRank-adversarial web pages isn't google's fault, exactly. The fact that they've been moving in the direction of more natural language-ish and less structured queries is their fault, but insofar as the right structured search for measles is "measles" instead of "measles -conspiracy -obama -mccarthy -autism" -- which seems right to me -- then what they're doing is sort of orthogonal to the changes in the kinds of queries they're built for.


Posted by: Sifu Tweety | Link to this comment | 07-23-15 6:45 AM
horizontal rule