These are all reasonable questions and things we wrestle with. Here's some of
my perspective. Others in the team are welcome to chime in with additional
thoughts.
On Mar 13, 2015, at 3:57 PM, Gervase Markham <[email protected]> wrote:
On 06/03/15 12:56, Hanno Schlichting wrote:
We've finally been able to announce our new partnership with Combain AS, head
over to our blog post:
https://blog.mozilla.org/services/2015/03/06/combain-deal-helps-expand-mozilla-location-service/
A commenter on that blog post asked the following question:
"I think the community deserves to know more details about this data
exchange. Is Mozilla's proprietary Wi-Fi database now the IP of Combain?
Our current database will not be going over to Combain. They will be receiving
new readings that we acquire. I don't think IP is the correct term to use here
- it implies it is now somehow exclusively theirs and they can keep others from
it. They merely have a copy of data we send to them.
Are stumbler contributions now being used for that company's profit?
Indirectly, yes. Ultimately, that's the goal of this project - to make the map itself
just a fundamental piece of data, not "owned". We believe that this will help
encourage an array of companies to sprout up to provide superior offerings and greater
choice in the geolocation space from what is available today.
Of course, to get there, we have a long way to go, both in terms of filling in
the map and working out the various privacy issues. Combain helps with both of
these goals, as well as helping us with improved lookups.
Since the Wi-Fi database was deemed too private to release to the
public, unlike like the cell ID database, what precautions are being
taken when sharing it with Combain?
Combain is, in many ways, the ideal initial partner. They are interested in the
data to improve their overall positioning map, and are using it in ways that
are relatively innocuous. The problem with the raw data is that it does have
sharing concerns that are still to be worked out, and Combain is already used
to these from maintaining their own raw-data database. They'll handle our data
as carefully as they handle their own.
For those who trust Mozilla, what grounds are there for extending that trust to
Combain?
You shouldn't, inherently. We have pretty good radar for how companies that want to use
the data plan to do so, and Combain "feels" pretty good to us, but there's
nothing contractual enforcing that. Their interests (which can be seen on their website)
are definitely in providing accurate positioning, rather than using the data in less
appealing ways, but ultimately location services are sold - via api - to third parties,
and it's not reasonable for us to be dictating what those third parties can or cannot do.
In the event that Combain fails to live up to our expectations (general, not
contractual), we have the option to terminate the contract any time.
What data from
Combain, if any, will ever find its way into the public domain?”
So I have to be a little handwavy here, because we're talking about a special
plan for Combain data. Let's talk about the generic case first.
Our general contract places no requirements on the data. It'll go into the main
corpus and, as we figure out how to share that, can be shared. Other contracts
may have restrictions on what we can do with the data; those contracts are
obviously less important to us, but if they can help improve MLS, we'll still
consider them.
The Combain contract is a general one. However, we've been wrestling with how to measure
the quality of MLS, and one of the things we're thinking of doing is moving the Combain
data into its own dataset and using it for testing. Our own stumbler data isn't good for
that - it's biased towards data that we already have - so this data represents a unique
opportunity. So the answer to this direct question is "no, not for now", but
not for the reasons that people might think. Ultimately, it should one day be included.
Hopefully these are useful answers. Eventually, we'd like to share as much as
we possibly can, and we're happy to talk to others who want to use the data to
see if we can find a reasonable way to make it happen, but there's a lot to be
solved before we get all the way there.
Toby
_______________________________________________
dev-geolocation mailing list
[email protected]
https://lists.mozilla.org/listinfo/dev-geolocation
