[Osmf-talk] Results of OSMF Member Vote

[Matija Nalis]

On Sun, Dec 27, 2009 at 06:01:24PM -0700, SteveC wrote:
> On Dec 27, 2009, at 5:33 PM, Matija Nalis wrote:
> > So, what I'm talking about is that while majority of OSM contributors might
> > not care if the license ends up to be CC-BY-SA or ODbL (which I fully expect
> > majority of them really won't care), they would care a PLENTY if due to such
> > low turnout *more than a half* of the data (45%+11%+xx%) and years of work
> > *gets deleted* in few months time. I know I would.
> 
> Do you have any idea how to make it better? We're going to have an

It depends what you mean by "it".

> extensive amount of work following up with all those that say no or don't
> respond and I'm actually quite hopeful that most people will click yes.

If by "it" you mean solving the "companies are afraid to use OSM maps
because we've chosen problematic license" problem, then yes, see below.

I am of opinion that Creative Commons were mostly right in deciding that CC0
(or other PD-alike) is the only way that will not lead to much unintended
pain and suffering [0].

We've (at least it seems to me) already discovered that CC-BY-SA is doing
more harm than good (as companies are afraid to use the OSM data). If we are
going to endure pain to amputate the ill leg, it might be wise to first
choose the leg that actually *is* ill, instead of making a wrong cut :-)

So I would say that a move to CC0 (instead of move to ODbL) would be much
more successful. If the unofficial polls are to be trusted [1], not only
would CC0 be more popular, but it would actually be in line with OSM stated
goal "OpenStreetMap creates and provides free geographic data such as street
maps to anyone who wants them. The project was started because most maps you
think of as free actually have legal or technical restrictions on their use,
holding back people from using them in creative, productive, or unexpected
ways", much more so than ODbL [2].

It seems to me people also wouldn't have such a need to make a fork(s).

But the most important part is there also wouldn't be a need to make soon
and abrupt cuts with deleting big chunks of data and alienating your users.
You could instead just make a combined map of relicensed and newly licenced
stuff under CC0, and leave the old CC-BY-SA stuff on the map under CC-BY-SA
license.

So with time (and we don't have a time limit here, much less such a short
one as proposed for ODbL) bigger and bigger parts of the map would become
copyright/database-unencumbered (under CC0), and small parts would still
remain under (dubious) CC-BY-SA, but still available to anybody [3] who is
not afraid (of being sued) to use them (for example everyone who does not
redistribute and is not bound by copyright, which might actually be a big
majority of private users).

You could even return (for API) or display (for web2.0 map) a license
depending on bbox being viewed/retrieved - either CC0 or CC-BY-SA, so the
company interested in using OSM could choose to use just CC0 parts.

One could even make a "CC-BY-SA" map (something like "NoNames" for broken
streets) so people with inclination can find and "clean" such areas to new
map.



However, if by "it" you mean "how to minimize the losses while moving to
ODbL, even if other license would have much lower losses; but we have other
interests in forcing ODbL", then I guess (if the transition period could be
prolonged by at least a year or few) the last two paragraphs above might
probably be adapted even to such scenario.

Additionally (and with much more technical work and/or resources) you could
give people tri-choice on relicensing (CC0, ODbL, and default if they do not
choose anything "CC-BY-SA") and make three maps from the same (sub)set(s) of
data - "completely free map, do what you want" one (CC0 only), "legally
correct copyleft map" (ODbL + CC0 combination released under "ODbL") and
"whole map, on your own risk if you redistribute" (CC0+ODbL+CC-BY-SA).
Handling the interdependencies when editing is left as an exercise to the
reader :-)

> That aside, the numbers are a lot better than you might think. Out of the
> 200,000 accounts, 70% have never edited anything and then those that have

That indeed sounds better than I thought. It escapes me why those 70%
bothered to create accounts if they never had a need to (ie. never edited
anything) ?

> edited are distributed in a scale-free way. Let's say 20% built 80% of the
> data, which I think is roughly correct then the key people are 200k * .3 *
> .2 or about 12,000 people. Taking a guess, if 90% of those click yes then

Hence my proposed poll for the OSMF members to define *how much* data loss
would be acceptable to the OSMF.

As you say, It doesn't matter at all if 60% of the OSM accounts don't
relicense to ODbL if *none* of that accounts never edited anything ! 
But even 20% of the OSM accounts that don't relicense would be terrible 
if they are that 20% that had built 80% of the map.

If 90% click yes, and we catch half of remaining 1200 people and get them to
click yes, that would be 5% loss (and acceptable to majority I would think,
me included).

if however we have results like OSMF poll (only 45% bothered to vote "yes"),
and add to that interdependencies problems, and we don't manage to convince
the rest to relicense; it would easily lead to, say, 70% data loss, which
might be much harder to accept for majority of members (I'm guessing again).

> we only are going to have problems with talking to 1,200 people which is
> entirely manageable. And better yet, we can prioritise our time talking to
> the ones who have edited the most.

Even so, after all hunting for people and talking and convincing them, at
one point we will have to stop and say "we've contacted and convinced
everyone we could, and with current situation, xx% of the map will be lost.
Do we proceed or not ?"

I'm only saying that that "xx%" should be put to vote and defined *now* in
advance. Would you accept such a vote ? (and why not if you wouldn't ?)

> So I think it's all achievable and if we just sat around talking about the
> doom scenarios we wouldn't have got as far as we have.

It it turns out (as you expect) with negligible losses, great for everybody
(me first!), and I won't be sad about any time "wasted" making backup plans
to handle less likely scenarios.

But if the "doom" really does come, then the "hah ! good thing we didn't
waste any time thinking about doom scenarios" won't be so appreciated :-)

I think "wasting" at least some time preparing even for situations one
considers less likely to happen is prudent thing to do. Of course we
shouldn't waste extreme amounts of effort on scenarios that have extremely
small chance of happening, but we should have *some* backup plans in case
things *do not* end up as we hope.

That is why one makes backups anyway, isn't it ? Hoping (s)he'll never need
to use them (but still "wasting" time every now and then to make them).

Matija

[0] http://wiki.openstreetmap.org/wiki/ODbL_comments_from_Creative_Commons

[1] http://doodle.com/feqszqirqqxi4r7w

[2] I'm afraid that a "license" that is actually a combination of copyright
    license, database rights license and contract aimed to be internationally 
    valid (especially one without any jurisdiction defined!) will hardly be 
    much more welcomed by many company lawyers (IME, many will not seriously 
    believe that such a beast could actually be construed)

[3] that would be everybody who are using the map *now*

-- 
Opinions above are GNU-copylefted.