The only way to know if these ua-profiles are new devices and not random noise 
is to somehow correlate them to real mobile web traffic. Thats going to be 
challenging for several reasons:

1) ua-profiles do not contain any user-agent (or identification) information. 
Infact, they seem to be getting more and more cryptic.
2) This project does not have a good source of mobile web traffic logs. I am 
not talking about getting logs from smaller websites, we need high traffic logs 
which attract a variety of devices. 

So maybe its better the other way around. Rather than have ua-profiles drive 
our device discovery, we should be using traffic logs to do it and then 
hopefully match the devices back to their profiles that way...

Regarding mobile devices, browsers, and OSes, yes, I totally agree. I think the 
time where devices came stocked with a static OS and browser is coming to an 
end. I dont think phones are going to be as flexible as desktops, but assuming 
OS and browser based on device is likely a misclassification.


________________________________
 From: eberhard speer jr. <[email protected]>
To: "'[email protected]'" <[email protected]> 
Sent: Friday, August 29, 2014 10:43 AM
Subject: Data
 

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Hi,

quick update on the analyses of 'harvested' 'new' UaProfiles.

Prelim. analyses turned up 2,048 *possible* new devices.
That number can only go down, but even if 10% comes thru further
testing...not bad.

I can't say for sure, but I have noticed at least some of the devices
Reza added to JIRA.

Other maybe important observation :

Some OEMs or rather UaProfile 'publishers' [fyi : most uaprofiles have
no copyright notice or Apache License] update the uaprofile of a
device model when the Android version changes.
This seems to suggest, quite reasonably, the upgrade of the OS on a
'same' affects more than just the os_version property...
I figure out [or try to] what if any those differences are and who and
if they are significant to us.

If anything it shows a shift away from the "Device" aspect to the "OS"
and "Browser" aspects.

esjr
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v2.0.22 (MingW32)

iQEcBAEBAgAGBQJUAJGqAAoJEOxywXcFLKYcemgH/3scdZobFwCSFlZc+FAivjhr
NeUKvMhvtpjjmrviQWRuUFFTiZUQxMP/gO3J2X+oHBqO8uLAfcshCGmb4O/Hr0NJ
aKz0H+PeGPACnymCIQkpLRNKQ3jDefgTLTOW+rDFCHgXLo8ppICqO/M7Z7BOwSGa
t12pAh52fFLMO1NyN6LE4t77CiNXnQ5n/SJ2EoFpStxTftOME3PGRHb9i3m1iGqD
BDP8T9GDJaqoIjqfQ5uin8btdPeFOW5XnuEXV0Q7uf7IRkPQ0DXtzPIHZ1pYDU7b
IO3Mi/Ihb+lvHIvUB3uA3uj/0Y/TvrlXNTB19/pGjuKLCjIXuo3NirfmDcgAdec=
=ktMb
-----END PGP SIGNATURE-----

Reply via email to