How Google Ambushed Microsoft and Changed the Subject | Hacker News Reader

How Google Ambushed Microsoft and Changed the Subject

techcrunch.com

32 pointsHardyLeung14 comments

>"I can’t condone any kind of plagiarism or cheating—and that is what Microsoft’s usage of Google data seems to amount to."

By academic standards, Google plagiarizes many times a day.

In Google's "sting" operation the description Google provides for www.cfacu.org is copied directly from DMOZ. Academic standards don't cut any slack for the ODP license and similar terms.

See:

http://googleblog.blogspot.com/2011/02/microsofts-bing-uses-...

and

http://www.dmoz.org/search?q=clyde+findley+area+credit+union

The best argument Microsoft has is: "You Too". Basically: "What I did wasn't wrong because the person complaining does it too."

My comment was not a defense of Bing but an attempt to point out the absurdity of applying academic standards Microsoft's conduct. Were one to apply academic standards to Binggate, it seems somewhat probable that Google's experimental results would have difficulty passing peer review.

Interestingly from a plagiarism standpoint, the description returned by Bing for cfacu.org is far better cited, since it is directly from the website itself rather than from an entirely uncited site like Google's.

In terms of academic standards, the results returned by Bing would probably not to be categorized as plagiarism but rather as not meeting style guidelines [such as APA format].

In this case, I think "You Too" is a pretty good argument.

By the standards of web search, pioneered by Google, bootstrapping rankings from clicktrails fits right in with lots of other gray, indirect reuse of others' editorial judgement. The de facto standards here are looser than standards in traditional fields, and that's a good thing!

I want Google to be able to liberally use text snippets from around the web. I want web crawling to be allowed by default. (You have to take affirmative steps, adding a robots.txt or other directives, to opt-out. No other written material is treated the same way!) I want most incidental 'infringement' from automated or user-driven processes liberally forgiven. I want Google and all its competitors to use clicktrails and similar signals, if legally available, to improve rankings and fight search manipulation.

As much as such practices offend copyright maximalists – a category Google has rarely been included within – these practices also create a giant bonanza of public benefit.

you don't have to be circumspect about it, it's also the description google uses if you search directly for cfacu. yahoo also uses it, and it's in several business directories, which you can see by searching on that phrase.

can you actually source it from DMOZ though? if so, they should call them on it, the open license seems to require attribution.

edit: I see google says they use DMOZ to populate descriptions if they don't have one: http://www.google.com/support/webmasters/bin/answer.py?hl=en...

yeah, sorry, I don't find Bing's behavior particularly egregious, but I'm finding equating it to compliantly using open material about as compelling as the arguments that are forgetting that robots.txt exists.

Most of Wadhwa's essays just get a "meh" from me. Nothing particularly original, not always pointed, and they usually don't leave me fulfilled. Perhaps this one may have been useful if the reader had been living under a rock for the last week, but those people don't frequent tech blogs.

If Blekko sticks to their word, I'd be inclined to give them a real chance. Sick of my search results being padded with spammy results.

EDIT: Having looked at them, I'm a little unclear about what they're trying to do. Are they actually letting end-users tag the web? I.e., Google not allowing users to block certain sources.

Yes, we're letting end-users tag the web. Try this video: http://vimeo.com/14593120 ... for the spam example, you can mark a site as spam with a single click. That blocks it only for you, but occasionally we look at aggregate data, and that's why we dropped those 20 sites from our index: because our users don't want to see them.

That is great to hear! I'm going to put it through its paces! I'd love to share feedback with you once I do.

>"We need a standard measure of web quality. "

To think that you can come up for an objective standard of quality for any search term is ridiculous. Why not just have a search engine return whatever your objective standard demands everytime and have a perfect score, if it's so easy?

A metric that can measure the quality of results but not produce them is entirely plausible. Many problems can be verified much more easily than they are solved.

For example, you can measure how often results are clicked but there is no straightforward algorithm to generate click-prone results.

Many articles on TechCrunch seem to be the gossipy type and are not much better than articles in a content farm that the author is railing on.

Humans like to gossip. That's why almost all reporting is boiled down to variations on horse race calling and theatre critique.

Reply on news.ycombinator.com