User-agent string changes

Re:User-agent string changes by molf on Saturday November 22, 2014 @11:17AM (#8645845)

  Mozilla/5.0 (Windows NT 6.4; WOW64)
  AppleWebKit/537.36 (KHTML, like Gecko)
  Chrome/36.0.1985.143 Safari/537.36 Edge/12.0

This is getting ridiculous. Every browser is now pretending to be all browsers at the same time, and a simple version change apparently warrants an entire article.

Not sending the UA string would avoid this madness. But on the other hand a UA string is great for analytics. Is it possible to come up with some method that discloses information about the user agent which can be used for analytical purposes but cannot be used to change the bahaviour of a web page?

[ Parent | on HN ]

Re:User-agent string changes by userbinator on Saturday November 22, 2014 @11:54AM (#8645953)

Is it possible to come up with some method that discloses information about the user agent which can be used for analytical purposes but cannot be used to change the bahaviour of a web page?

It's called "browser fingerprinting": https://panopticlick.eff.org/

[ Parent | on HN ]

Re:User-agent string changes by ancarda on Saturday November 22, 2014 @12:26PM (#8646043)

It's useful to know the OS if you want to offer different binaries for each OS. Maybe just "BrowserName/Version (OSName/Version)" so

    Firefox/35 (Mac OS X/10.10.1)

[ Parent | on HN ]

Re:User-agent string changes by tacostakohashi on Saturday November 22, 2014 @01:12PM (#8646162)

One approach would be for browsers to send a string that is along the lines of a uuid generated at compile time, or an MD5 of the executable.

That way, analytics vendors or libraries could maintain a database that maps IDs to specific browsers releases, platforms, features, etc if they care about those things enough, but short-sighted web admins would have no hope of doing anything with primitive regex matching of the UA string.

[ Parent | on HN ]

Re:User-agent string changes by tracker1 on Sunday November 23, 2014 @04:07AM (#8648230)

You're welcome to do just that.. md5(useragent) and keep your own definitions... be mindful that in IE in particular plugins affect the string itself. :-(

[ Parent | on HN ]

Re:User-agent string changes by firepacket on Saturday November 22, 2014 @05:36PM (#8646912)

You mean a super cookie?

Not only would this violate tracking laws in some areas, but who would ever want to use such a thing?

[ Parent | on HN ]

Re:User-agent string changes by tacostakohashi on Saturday November 22, 2014 @08:05PM (#8647378)

No. Generated at the time the browser executable is built (by the vendor, linux distro, etc), and hence shared by all the people using that browser executable. So, only good for identifying the browser, platform, OS and features, and much less so the users.

[ Parent | on HN ]

Re:User-agent string changes by pbhjpbhj on Saturday November 22, 2014 @11:26AM (#8645871)

Isn't this a trademark infringement? They're using AppleWebKit, Chrome, Safari to identify their product when those terms are protected from use via trademark law.

I know this isn't a new thing per se but this seems to be an especially egregious misrepresentation.

[ Parent | on HN ]

Re:User-agent string changes by lambda on Saturday November 22, 2014 @01:36PM (#8646237)

No, because this is not in any way going to cause consumer confusion, but instead is simply used for compatibility purposes.

[ Parent | on HN ]

Re:User-agent string changes by pbhjpbhj on Saturday November 22, 2014 @04:17PM (#8646690)

Confusion is the measure used when trademarks aren't registered and is much harder to prove. When considering registered trademarks it's not necessary for confusion to be actually present at all - 100% of customers can know that your product isn't an iPad but if you advertise it as an iPad then you're infringing: it doesn't say "AppleWebKit compatible" in that string and the string is not required to make the browser function at all.

If, for example, people offer alternative services to people with an Apple UA string (BBC iPlayer did this) then you're impersonating such tech use via the trademark - it might be now for example that people need to turn off specific iPad enhancements as they break IE11 - that's commercial interference using a trademark. The trademark isn't needed in any technical way for IE to operate as it should, so it's not a compatibility issue.

Website content owners can choose to cut out certain browsers if they wish, impersonating another browser is fine .. but if you do that as a company utilising someone else's trademark to do so .. it just has feeling for me of not sitting quite right with what I know of TM law.

[ Parent | on HN ]

Re:User-agent string changes by firepacket on Saturday November 22, 2014 @05:32PM (#8646900)

I would think the courts would recognize a difference between advertising to the end user for the purpose of purchase/consumption, and advertising to 3rd parties as part of a compatibility protocol.

The UA string is not visible to the consumer, and wouldn't affect the number of people who download and install the browser.

Its only purpose is to ensure compatibility with the server after the browser has already been installed.

Also I wouldn't say the browser is actually impersonating another browser. If it was trying to impersonate, the UA string would need to match the other browser exactly and it doesn't attempt to do that. The string is still uniquely identifiable as an IE11 UA.

[ Parent | on HN ]

Re:User-agent string changes by freehunter on Saturday November 22, 2014 @11:34AM (#8645890)

Is it any more of a trademark infringement than 'IBM PC Compatible'?

[ Parent | on HN ]

Re:User-agent string changes by cma on Saturday November 22, 2014 @04:08PM (#8646670)

Sega and/or Nintendo tried to require games to display their trademarks before their system would run them, courts ruled that third parties could use the trademarks to overcome such systems.

[ Parent | on HN ]

Re:User-agent string changes by Dylan16807 on Saturday November 22, 2014 @11:35AM (#8645893)

Will anyone be tricked? No. So not infringement.

[ Parent | on HN ]

Re:User-agent string changes by minot on Saturday November 22, 2014 @11:34AM (#8645887)

Infringement, as far as I know is a civil matter. Who is going to sue when everyone is doing it?

[ Parent | on HN ]

Re:User-agent string changes by mcbutterbunz on Saturday November 22, 2014 @11:53AM (#8645950)

I hope not. The last thing we need is a new round of lawsuits. Maybe everyone can agree on a set of rules for communicating this data.

[ Parent | on HN ]