Establish criteria for selecting a region

[DouglasSherk]

We're lacking criteria for actually selecting a region. I think we punted this originally because we thought that researching it would help, but this is really something that the design challenge people have to provide, and they haven't.

Our one idea to go on was to compare existing systems in countries with serious ERW problems and compare them with countries that either used to have this problem or don't really but still have good systems for dealing with them. This might give us some insights but I'm skeptical of the value of that.

Here are the factors that I think we should consider when making this decision, in descending order:

• How easy will it be to find people in Toronto who came from this region?
• How serious is the problem there? How many people in 1000 in that country die every year from ERW detonations?
• Is there enough infrastructure and phone penetration for us to even be able to deploy there?
• Does Mozilla UR have data on this region?
• Is there currently a war going on in that region?
• Is their culture similar enough to ours that we won't run into serious problems at every turn due to iconography, customs, established patterns, etc. that we aren't aware of?

[vtsatskin]

I agree that we need to address this before we can evaluate and select regions.

I suspect it's going to be hard for us to narrow down what regions we want to focus, as I was overwhelmed and surprised with the near number of countries and regions affected that you outlined in #12. I think reaching out to the organizers of the design challenge to see if they have a list of regions they think are important to focus on would be a good thing to do. See #23. (As an side, I'm curious what their evaluation criteria is. Is it political, humanitarian or something else?)

Comparing systems in regions with serious vs non-serious ERW problems was to gain insight into systems that are effective vs systems that are not. The assumption was that seriousness of ERW problems are inversely correlated to effectiveness of their existing ERW system. So I agree that there is not much value in doing that method to help select which regions we focus on.

I like the idea of making a list of factors, and ordering them by priority. If we agree on this, we can make evaluating easier. Overall, I agree with your list. How come you rank finding people in Toronto as higher than how serious the problem is?

At the end of this research phase, do we want to narrow it down to a single region or perhaps a few to focus on?

I think we should aim to have a few regions to help us make a generalizable and/or customizable solution for each region. Generalizable as in the solution is not too specific for a region (i.e. using region specific systems, perhaps some custom mine reporting software?). Customizable as in we can swap out locales so that strings and iconography can be customized. My worry is that if we only choose one, we won't know what different requirements are. However, choosing one region will allow us to focus our efforts.

[DouglasSherk]

Comparing systems in regions with serious vs non-serious ERW problems was to gain insight into systems that are effective vs systems that are not. The assumption was that seriousness of ERW problems are inversely correlated to effectiveness of their existing ERW system. So I agree that there is not much value in doing that method to help select which regions we focus on.

Okay, I must have missed that at first. I don't think that's a good assumption since there are so many other factors that could play into this.

I like the idea of making a list of factors, and ordering them by priority. If we agree on this, we can make evaluating easier. Overall, I agree with your list. How come you rank finding people in Toronto as higher than how serious the problem is?

I don't think they actually care if we choose the region with the absolute worst problem overall as long as we choose one that has a serious problem, and we do a good job at solving it. I think testing is going to be important for solving it well. So I think that we should prioritize having people to test it on, and all data available that we need, above everything else.

I think we should aim to have a few regions to help us make a generalizable and/or customizable solution for each region. Generalizable as in the solution is not too specific for a region (i.e. using region specific systems, perhaps some custom mine reporting software?). Customizable as in we can swap out locales so that strings and iconography can be customized. My worry is that if we only choose one, we won't know what different requirements are. However, choosing one region will allow us to focus our efforts.

Yeah, that's a good point. Personally, I think we should pick one region and try to solve the problem really well there. We can pick a second region which we use to evaluate every decision against, and make sure that our solution is fairly generalizable for both of them. That way, we're not locking ourselves into that one region, but we're not slowing ourselves down significantly or making the scope too large.