It's a little out of the ordinary for the content we post here, but one of the things that we at 3PO-Labs (and specifically Eric) find ourselves doing with some frequency is advocating on behalf of other developers who end up in situations they don't know how to resolve. Often this happens during the certification step, where a first rejection can seem like an insurmountable obstacle, especially for folks who are less familiar with how things work behind the scenes than we are. We recently took some time to argue on behalf of a few folks, and I wanted to share what that looks like a bit more broadly...
Note: Shortly after we wrote this article, the Alexa State of the Union session at AWS re:Invent announced new monetization options, and also specifically called out the ability to do donations to charities via Amazon Pay. This furthers our belief that the purpose of the rule was not to prevent donations in general, but rather to prevent skills from asking for money for themselves.
The issue we were seeing this time from a few different sources was that the Alexa cert team was taking a stance on one of their policy guidelines - specifically around solicitation of "donations" - that was a little bit pedantic and not really in line with what (we believe) the rule was meant to solve. In this case, our approach was to reach out to folks in cert and do what is more or less the equivalent of filing an amicus curiae. We wanted to pass along the (very lightly edited) email we sent, and see whether or not you all agree about the arguments made:
To Whom It May Concern,
I wanted to reach out about some rumblings I've been hearing again lately among the dev community about a weird thing that's happening with cert. It doesn't pertain to my skills, so I have no horse in the race, but it's one of those cases where I'm seeing other devs throwing their hands in the air and giving up because they believe it's a lost cause, and it bothers me to see the platform lose solid contributors for bad reasons. I'm hoping you can pass this along to whoever else might be involved in the decision making process for cert guidelines.
Essentially, it seems your reviewers have been keying in on the word "donate" and broadly applying a rule that seems to have been added for a very specific use case. The line in question is policy guideline 4.C:
"c. Solicits donations from end users."
So, yes, the rule says "soliciting donations" isn't allowed, but the context here is pretty obvious. This rule seems to have been written to prevent donations in support of the skill itself, which would more or less end up as a way of circumventing the pay-to-play monetization guidelines for skills. And, in my opinion, the rule makes perfect sense through that lens.
The problem is that there are use cases where a donation solicitation may occur that have nothing to do with the skill itself being monetized. The case that I'm hearing lately is one where folks in the dev community are building skills for non-profits. Cert seems to be rejecting skills for non-profits that solicit donations to the non-profit, even though there is no kickback to the skill itself.
The reason I believe this rule is incorrectly applied is that the donations have nothing to do with skill monetization. Rather, it's a transaction between a third party and a user, taking place in an external system, which is something you already allow (see: Uber, Dominos, etc). And I think the hangup here is really the word "donate". A "donation" to a non-profit is much closer to the purchase of a pizza than it is to a donation for a skill you really like. The only difference between giving your money to Dominos and giving your money to a non-profit like Meals on Wheels is that in the former, you get food delivered, whereas in the latter, someone else gets food delivered. There's not really a clear ethical or business impetus to draw a line in the sand between the two, which is why I believe that this is just an oversight of a set of rules that are still evolving.
As such, I think you could clarify section 4 of the policy guideline rules to look something like as follows:
"4. Purchasing and Currencies
a. Offers a separate skill store or recommends other skills.
b. Offers compensation for enabling other skills, or for providing product reviews on skills. Examples of compensation may include digital goods, physical goods, or monetary transactions.
c. Solicits donations from end users in support of the skill.
c.1. Promises or provides additional skill functionality in exchange for donations to a third party."
This way users could still complete transactions with the Red Cross or whatever, but you'd also still prevent people from circumventing the skill monetization rules.
Additionally, you'd have the benefit of donation solicitations occurring only in skills (that users have chosen to enable, mind you) where solicitations might be expected. There's certainly an argument to be made that if one of my skills like CompliBot started asking for donations, that would be annoying to the users, and look bad for the platform as a whole. But at the same time, if a person enables a skill for Save the Whales, their expectation is almost certainly going to be that they will be afforded an opportunity to help save whales via monetary contribution. The fact that a policy-level guideline prevents this is actually actively making the platform counter-intuitive for users.
And finally, I'd appeal to the futility argument. Right now, one of these two statements will fail your cert pass, and the other will not:
1. Reserve your name's slot now! Buy a slot on our page, spending as little or as much as you'd like.
2. Donate now! Donate as much or as little as you'd like - all backers get their name listed on our backers page.
These two are functionally equivalent, but the first one doesn't "solicit a donation". In this case, the policy hasn't caused a user to have a different experience, they are still presented with the same option. Instead, it just forced the skill developer to use verbal gymnastics to get around a rule that wasn't meant for them.
I hope this all makes sense. I've obviously made some assumptions here about what you guys were trying to optimize FOR with rule 4.c., but I think there's a pretty solid business case for you guys to update that to be less broad.
Regards,
Eric
I wanted to reach out about some rumblings I've been hearing again lately among the dev community about a weird thing that's happening with cert. It doesn't pertain to my skills, so I have no horse in the race, but it's one of those cases where I'm seeing other devs throwing their hands in the air and giving up because they believe it's a lost cause, and it bothers me to see the platform lose solid contributors for bad reasons. I'm hoping you can pass this along to whoever else might be involved in the decision making process for cert guidelines.
Essentially, it seems your reviewers have been keying in on the word "donate" and broadly applying a rule that seems to have been added for a very specific use case. The line in question is policy guideline 4.C:
"c. Solicits donations from end users."
So, yes, the rule says "soliciting donations" isn't allowed, but the context here is pretty obvious. This rule seems to have been written to prevent donations in support of the skill itself, which would more or less end up as a way of circumventing the pay-to-play monetization guidelines for skills. And, in my opinion, the rule makes perfect sense through that lens.
The problem is that there are use cases where a donation solicitation may occur that have nothing to do with the skill itself being monetized. The case that I'm hearing lately is one where folks in the dev community are building skills for non-profits. Cert seems to be rejecting skills for non-profits that solicit donations to the non-profit, even though there is no kickback to the skill itself.
The reason I believe this rule is incorrectly applied is that the donations have nothing to do with skill monetization. Rather, it's a transaction between a third party and a user, taking place in an external system, which is something you already allow (see: Uber, Dominos, etc). And I think the hangup here is really the word "donate". A "donation" to a non-profit is much closer to the purchase of a pizza than it is to a donation for a skill you really like. The only difference between giving your money to Dominos and giving your money to a non-profit like Meals on Wheels is that in the former, you get food delivered, whereas in the latter, someone else gets food delivered. There's not really a clear ethical or business impetus to draw a line in the sand between the two, which is why I believe that this is just an oversight of a set of rules that are still evolving.
As such, I think you could clarify section 4 of the policy guideline rules to look something like as follows:
"4. Purchasing and Currencies
a. Offers a separate skill store or recommends other skills.
b. Offers compensation for enabling other skills, or for providing product reviews on skills. Examples of compensation may include digital goods, physical goods, or monetary transactions.
c. Solicits donations from end users in support of the skill.
c.1. Promises or provides additional skill functionality in exchange for donations to a third party."
This way users could still complete transactions with the Red Cross or whatever, but you'd also still prevent people from circumventing the skill monetization rules.
Additionally, you'd have the benefit of donation solicitations occurring only in skills (that users have chosen to enable, mind you) where solicitations might be expected. There's certainly an argument to be made that if one of my skills like CompliBot started asking for donations, that would be annoying to the users, and look bad for the platform as a whole. But at the same time, if a person enables a skill for Save the Whales, their expectation is almost certainly going to be that they will be afforded an opportunity to help save whales via monetary contribution. The fact that a policy-level guideline prevents this is actually actively making the platform counter-intuitive for users.
And finally, I'd appeal to the futility argument. Right now, one of these two statements will fail your cert pass, and the other will not:
1. Reserve your name's slot now! Buy a slot on our page, spending as little or as much as you'd like.
2. Donate now! Donate as much or as little as you'd like - all backers get their name listed on our backers page.
These two are functionally equivalent, but the first one doesn't "solicit a donation". In this case, the policy hasn't caused a user to have a different experience, they are still presented with the same option. Instead, it just forced the skill developer to use verbal gymnastics to get around a rule that wasn't meant for them.
I hope this all makes sense. I've obviously made some assumptions here about what you guys were trying to optimize FOR with rule 4.c., but I think there's a pretty solid business case for you guys to update that to be less broad.
Regards,
Eric
Thoughts? Good approach/bad approach? Should that rule be changed? If you're running into this type of issue with cert, we'd love to help you if we're able. A lot of times it's as simple as changing the wording you use in your testing notes so that the reviewers don't focus on the wrong things.
RSS Feed