Fedora and Google search: the final report

[Bryan Che]

I think suggestions like this would be great.  Once I get a SIG Wiki 
created, we'll need to figure out how to allow projects to join and get 
cycles.

"Trellis" is a cool name, but I'm also wondering if it would be useful 
to have a more transparent name.  For example, it would be easier to 
tell people about the project and get them to donate cycles to it if the 
name gave more indication upfront about what the project was (that's why 
I used "Fedora at home" in my initial e-mail--so people would instantly 
understand what I was proposing).  The alternative is that we'd have to 
put a lot of marketing effort behind making people aware of a name like 
"Trellis."

Bryan

Paul W. Frields wrote:
> On Tue, 2008-03-11 at 13:18 -0400, Greg DeKoenigsberg wrote:
>> This represents my final report on this matter.  My conclusion: it's not 
>> practical to pursue a relationship between Fedora and Google at this time, 
>> for the following reasons:
>>
>> 1. RED HAT BEARS THE LEGAL RISKS FOR FEDORA.
>>
>> As much as Red Hat has worked to create an independent governance model 
>> for the Fedora Project, in the final analysis, Fedora is a property of Red 
>> Hat.  This means that any deal made between Fedora and a third party 
>> *must* be agreed upon by Red Hat legal.
>>
>> 2. THE BOILERPLATE CONTACT FOR GOOGLE'S CUSTOM SEARCH ENGINE (CSE) IS TOO 
>> RISKY FOR RED HAT LEGAL TO ACCEPT.
>>
>> The terms and conditions for the Google CSE program 
>> (http://www.google.com/coop/docs/cse/tos.html) are pretty onerous.  The 
>> biggest sticking point for Red Hat lawyers is Section 5, in which Google 
>> demands unlimited indemnification.  Negotiations to resolve this clause 
>> have been unsuccessful.
>>
>> 3. MECHANISMS LIKE THE FIREFOX SEARCH BAR ARE SPECIFICALLY EXCLUDED FROM 
>> THE GOOGLE CSE.
>>
>> It seems like there are two potential sources of revenue for Fedora in 
>> Firefox: the start page (which would probably be low dollar) and the 
>> search bar (which would probably be high dollar).  In actuality, though, 
>> the search bar is *specifically* excluded from the Google CSE agreement, 
>> in section 1.4, Appropriate Conduct.  Quote: "You shall not, and shall not 
>> allow any third party to: ... (f) directly or indirectly access, launmch 
>> and/or activate the Service through or from, or otherwise incorporate the 
>> Service in, any Web site or other means other than the Site, and then only 
>> to the extent expressly permitted herein."  Which means that unless we 
>> negotiate a deal with Google directly, we can't use the search bar to 
>> generate revenue at all.
>>
>> So that's it.  There are no simple avenues to pursue at this point, which 
>> means that we are now open to pursue other options (wikia, fedorasearch 
>> being the two I like best).  We should also consider whether we want to 
>> change the default search to use these, which might require a break from 
>> the Firefox brand.
> 
> I like the fact that wikia is all about transparent, open services.
> That makes them more like us than Google is.  They're also based around
> a community contribution model -- people donate spare bandwidth and
> cycles to webcrawl for indexing purposes.  I don't think wikia is
> competitive yet as a search tool from the user's point of view, but we
> certainly are in a position to help make them more so.
> 
> What if, as part of this new "community grid" project (which I like
> calling "Trellis"), we offered people the ability to pitch in with the
> web-crawling duties?  This is not limited to just Fedora users, but
> here's one possibility:  Installing user sees firstboot module providing
> Trellis.  User opts in, then from a menu of system resource donations,
> selects either (1) "wikia," or (2) "Fedora, I trust you, use this time
> for whatever you like," which maps to a certain proporation of donation
> for wikia (published somewhere, with a link).
> 
> This isn't purely about CPU cycles anymore, but it's still worth
> considering.
>