Open domenic opened 5 years ago
WhatsApp's social graph is pretty much entirely based off the users address book on device - you can't add someone properly without there being in your on device address book.
Emerging market use-cases are frequently that the apps and sites such as fb and instagram are now on the web and want to offer the same experience to users that don't ever install the app directly but only use the site.
It's good to hear we have one precedent in WhatsApp. That implies that in areas with high WhatsApp usage, the on-device contacts database is indeed likely to be useful. It would still be good to see some numbers though.
You seem to have misunderstood my concern about other social graphs. I was concerned that EM chat app X would prefer to connect (via OAuth) to e.g. Facebook, and get the user's contacts from that social graph, instead of getting them from the user's local device. You have one counterexample, which is good, but unless we're designing this API specifically for WhatsApp and others that use that same model (which might be fine!) I think it's worthwhile doing research on how many EM apps bootstrap their social graph via on-device vs. Facebook vs. local social networks vs....
@domenic I can think of several use cases.
I understand that it's possible to conceive of cases where this API is useful. What I'm asking for is evidence that this API matches the usage patterns of EM users in the wild.
Gotcha. Not sure what qualifies as evidence but all of the above stem from the challenges I faced in my experience building Flipkart Lite & Instagram Lite.
Ah cool, so the above was based on anecdotal evidence at least; that's helpful. (You phrased it in a way that made it sound like it was just theorizing.) And indeed, anecdotal evidence might be the best we can get.
The ideal form of evidence would be some numbers. For example, if there were a native app in EM that offers to get your contacts either from a social graph provider or from the native contact store, we could ask that app for statistics on how often each option is chosen. Or, a survey of the top N EM apps, and a breakdown of how many try to access the native contact store, vs. how many access social graph contact stores. Those were the kind of things I opened this issue hoping someone would do research on.
The blink-dev I2I for this feature motivates it by appealing to emerging market use cases. However, it's easy to imagine that such users avoid the built-in Android contacts app, in favor of social graphs through third party services such as Facebook, WhatsApp, WeChat, etc.
Do we have evidence that this API will actually be useful for a significant fraction of EM users?