app-mining

app-mining copied to clipboard

What is the problem you are seeing? Please describe. Apps are rewarded that are not used by many users

How is this problem misaligned with goals of app mining? This discourage other app publishers to develop high quality apps

What is the explicit recommendation you’re looking to propose? Add an app reviewer that analyzes and rates the usage of apps.

Additional context http://theblockstats.com/ shows number of users for multi-player apps

Projects about decentralized analytics were mentioned

Examples:

Name | App Mining Rank | Number of users

Gladys | rank 4 | I guess very few. Graphite Publishing wrong data Xordrive wrong data

Are there other examples?

@friedger I think the number of mulit-players you mentioned for Graphite Publishing and Xordrive are incorrect, from what I can see in http://theblockstats.com/

Also, Gladys uses single-player auth, so new sign ins won't show up here. The 4 is probably just from when the developer was testing it.

@hstove @muneebm thank you for the verification.

There is indeed no data for Gladys. However, there is only a potential of 35k users altogether vs 4billion on the web. Furthermore, the advantage of blockstack for Gladys users is not immediate, only in case they want to reset or replace a device. This is very comparable to OI ConvertCSV while OI ConvertCSV has a potential of 100k+ users (depending on what you count..)

I'm not sure how single player apps will perform here. eg SpringRole has atleast 1100 blockstack connects but we will never show up in blockstats because it can only track multi-player ones.

@kar2905 I don't promote using the multi-player data. I am very much for privacy and very much like apps that do not publish in the profile.

This is a place to discuss the aspect of app usage in the app mining algorithms.

@friedger yet your proposal only discusses a stat mechanism that excludes those apps you praise for privacy. -1 for theblockstats being mentioned anywhere near app-mining algorithms.

It seems like the Awario proposal would be a reasonable stand-in for real-life user tracking.

Just updated ticket title to reflect a proposal about http://theblockstats.com/ specifically. There are a few other potential methods but they will likely require PBC build some software. Would be great to get a discussion going specifically re installs and if we can overcome the single player challenge. cc @GinaAbrams

I don't think number of installs is a good measure similar to number of downloads in an app store. Good apps in the sense of the app mining criteria do not need many users.

The number only tells how many people have seen the app at least once. It does not say much about that app, maybe that the app is already around for a long time, maybe that lots of users tried it only on one specific date.

I would value privacy higher than user tracking/analytics.

Add an app reviewer that analyzes and rates the usage of apps.

Cool, I agree this is a great idea. If not Blockstats, then who/what? I value the discussion, just trying to get to a really specific action we can implement now.

I think we could probably use the blockstats as a stopgap for several months until we find a way to measure ongoing usage of apps (if that makes sense to even do). Using the blockstats doesn’t have to be long term, and doesn’t have to simply count absolute # of installs.

For example you could weight monthly growth in relation to other apps like 70% and weight total installs like 30% or something.

Also you could use the blockstats but weight names registered that cost money some proportion more, so you have a tax to Sybil attacks. Very interested ideas can emerge from this one.

Further the system could also give an outsized reward to apps that create IDs that visit more apps. So if Graphite creates an ID, they get +1 but if that ID then uses a second app +2.

Lastly, @dantrevino is it not simple good hygiene to have multiple identities? The onus is on the individual to have a burner ID and that seems perfectly reasonable for this purpose cc @jcnelson who agrees with this approach. I don’t have one bitcoin wallet and I don’t have one identity for the sake of privacy and security.

And single user apps are always ranked higher than multi user apps.

On Sat, 23 Mar 2019, 23:48 Patrick Stanley, [email protected] wrote:

I think we could probably use the blockstats as a stopgap for several months until we find a way to measure ongoing usage of apps (if that makes sense to even do). Using the blockstats doesn’t have to be long term, and doesn’t have to simply count absolute # of installs.

For example you could weight monthly growth in relation to other apps like 70% and weight total installs like 30% or something.

Also you could use the blockstats but weight names registered that cost money some proportion more, so you have a tax to Sybil attacks. Very interested ideas can emerge from this one.

Further the system could also give an outsized reward to apps that create IDs that visit more apps. So if Graphite creates an ID, they get +1 but if that ID then uses a second app +2.

— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/blockstack/app-mining/issues/67#issuecomment-475911342, or mute the thread https://github.com/notifications/unsubscribe-auth/ABYcWQoUUuTnOc6VHeV-B7pfFpmk_nNCks5vZq8sgaJpZM4b-yom .

In general, having a "stopgap" reviewer that looks at growth metrics makes a lot of sense.

Tracking growth in a decentralized way is a fairly complicated problem given user privacy but it's likely that Blockstack makes more progress on "privacy-focused growth tracking" in the coming months. As more sophisticated ways to track user engagement becomes available, the growth reviewer can also evolve with time.

Lastly, @dantrevino is it not simple good hygiene to have multiple identities? The onus is on the individual to have a burner ID and that seems perfectly reasonable for this purpose cc @jcnelson who agrees with this approach. I don’t have one bitcoin wallet and I don’t have one identity for the sake of privacy and security.

Its not a matter of using a burner id to not be identified. How are you going to count non-social apps?

Another issue, how many of my 12 burner ids are going to count towards SocialDapp "+1"s? i.e. If I use burnerA.id for my meetup group sharing, and burnerB.id for my hackathon planning, and burnerC.id for posting my Anti-Google manifesto, does SocialDapp now get 3 +1s, even though there is only one "user"?

Sounds like we have some major issues with theblockstats. Blockstack team is going to explore some more anti-fragile solutions this quarter. Iceboxing for now, will come back with new proposals later.