Task Prio Notes:

• Does not affect end-users \ production (as the investigation results do not directly affect users, but the engineers who will have to implement any compatibility fixes)
• Does affect development efforts
• Does not onboarding efforts
• Affects additional stakeholders (as this feature will be prepared and deployed by other teams in the movement)

Story Writing Notes:

• Add explanation on IP masking from metawiki.

Some thoughts I had during the wmhack session (T332079):

Two things that directly comes to my mind here there this might cause trouble is the link item feature (linking Wikidata items from the client) and Wikibase client's bridge. Link item is supposed to only be accessible to logged in users, both tools will need to "log in" on Wikidata (both using mediawiki.ForeignApi).

Especially interesting case here: User has no (temporary) account yet, uses one of the tools (shall we allow that?)… and gets a temporary account on Wikidata (does that work, will it (=the account login) propagate to the client wiki?).

I am monitoring this task as I would like to understand the consequences of IP Masking for analytics.

Timebox 16 hrs broken into

• 8 hr - Investigate places where Wikibase distinguishes between anonymous and registered users
• 8 hr - Investigate behavior of temporary accounts with cross-wiki API actions

Very related and probably also important for Product sense-making around this: T337103: Decide a standard approach for classifying temporary, IP and registered users. cc @Arian_Bozorg @Lydia_Pintscher

See the child-tasks T343799 and T343800 for the results. Please take a look and let us know if you're missing something or need the information presented differently or anything.

Also, the upstream implementations seem to still see very active development, so it might make sense to wait for them to settle somewhat and not rush our development:

• T336187: [S] Investigate: Creating temp user on non-EditPage actions
• T337103: Decide a standard approach for classifying temporary, IP and registered users
• T325592: Start a technical documentation page for IP Masking

(this task should probably move on to tech-verification after product is happy)

Thank you! Looking good. Moving this to tech verification.

Results from the subtasks copied with slightly adjusted language to: https://wikitech.wikimedia.org/wiki/WMDE/Wikidata/Reports/2023/2023-09-18_Ramifications_of_IP_masking_on_Wikidata_related_extensions

@Lucas_Werkmeister_WMDE please let me know what you think of the overall report and my adaption of your parts in particular.

FTR: That writeup took another two hours.

Looks great to me, thank you! (I’ve made a few small edits to the page.)

The edits look good to me, thanks!

If I understand IP masking correctly, the goal is to get rid of IPs as public identifiers. Instead we will use cookie-based "temporary accounts" as public identifiers. These are independent of the IP. I have two questions about this:

1. This seems feasible for browser-based editing. But what about API-based editing? Can non-logged-in users currently edit using our APIs? (I hope not, as it does not sound like a good idea. xD) How will/should this work in the future?

2. For Wikidata Analytics I got the impression that we will still store IPs internally, so nothing should change from an analytics perspective. Is that correct?

In T328454#9199247, @Manuel wrote:

If I understand IP masking correctly, the goal is to get rid of IPs as public identifiers. Instead we will use cookie-based "temporary accounts" as public identifiers. These are independent of the IP. I have two questions about this:

1. This seems feasible for browser-based editing. But what about API-based editing? Can non-logged-in users currently edit using our APIs? (I hope not, as it does not sound like a good idea. xD) How will/should this work in the future?

Yes they can. The edits made from entity pages are made via API calls from javascript, that includes edits from users that are not logged in.
The details how that will work in the future will need to be figured out. I guess the API response will include some info about the created account?

2. For Wikidata Analytics I got the impression that we will still store IPs internally, so nothing should change from an analytics perspective. Is that correct?

We will still store the IPs internally, also for anti-abuse measures. With temporary accounts, I could see new types of queries coming the table of analytics with respect to the behavior of anonymous editors, now that we sometimes have a better way to associate edits with the same editor over time than relying on IPs.