On 2011-02-02 09:02, Nick Mathewson wrote:

...

On Tue, Jan 25, 2011 at 9:04 AM, Lucky Green <shamrock@cypherpunks.to> wrote:

...

...

On 2011-01-25 01:09, Erinn Clark wrote: [...] Erinn, My feedback is that the Tor Project really, really will want a written and published policy of how far back an OS is supposed to be supported. Otherwise, you will get to have this discussion every time a new OS version is released.

Industry standard for consumer software that goes into the far corners of the world is "current and previous major version", which has different meanings depending on OS.

Hi, Lucky! I generally agree with you, but I do want to point out a point that makes me balk at applying your experience without more consideration.

It seems to me that the "industry standard for computer consumer software" is chosen to serve the goals of the software industry: that is, to sell software to people who can pay for it. As a means to that end, commercial software enterprises sometimes additionally support marginal userbases (those who can't pay, those who can't pay much) in order to increase the install base of their software and provide positive network effects for their paying customers.

We're in a slightly different position: we're achieving our mission to the extent that lots of users -- whether they can pay or not! -- can use our software. Also, we want volunteers with free (to them) bandwidth to be able to run our server software on the hardware and software they have lying around.

I suspect that, when all is said and done, the set of operating systems that it makes sense to support in the commercial software industry will work out to be almost the same as the ones Tor want to support -- but the reasoning is a little different, since industry needs to optimize profit-per-developer-hour, and we're trying to optimize social-benefit-per-developer-hour.

Hmm. My emails on this topic must have been quite poorly written to result in your feedback about them focusing on what I had intended to be little more than a footnote or parenthetical at the very end of the communication thread. I am aware of and concur that Tor's success metrics are different from those of most commercial software vendors. My analysis was entirely focused on helping Tor devise the best OS version support policy possible while assuming Tor's goal of "social-benefit-per-developer-hour". My bad if that was not clear from my writings. Though I am not quite certain what the best path would have been for me to take: omit any references to commercial software development best practices, even in annexes and footnotes, when discussing open source software? Somehow that doesn't strike me as the optimal answer. But be that as it may, I assure you wholeheartedly that my analysis was solely intended to guide Tor and its ecosystem, not that of commercial software development. Either way, even more core to my advice than which particular OS versions to support (the one area where any dichotomy between open source and the commercial world may or may not come in) was my point that Tor needs a written and published OS support policy, agreed-upon by the Tor team. Whichever version support policy the Tor team may arrive at and for whatever reasons. And whether or not the Tor team takes my advice as to which particular versions to support. --Lucky