Talk:Abstract Wikipedia/Updates/2022-07-15

Potential Analysis

I think that the Potential out of Abstract Wikipedia is propably much lower as expected as of today. In the past I heard from differnt examples with too optimistic estimations. At least contributing to Abstract Wikipedia and creating own templates is complicated and it will be propably for a long time. This is not a problem as it is not only about the amount of people what impact Abstract Wikipedia will have and I am interested in the technic behind and will maybe use it for other cases too. I think that will be not more than 200 regular active Volunteers at Wikifunctions. So it will be from my point of view a small Wikimedia Project and a powerful one. Because of the abstract way of writing things. If in the texts the most important facts are missing it is less inviting for people to contribute and so propably the number of people who will start contributing after reading an abstract article is low. Hogü-456 (talk) 21:58, 15 July 2022 (UTC)

I very much hope and expect that your prediction of 200 active contributors to Wikifunctions is too low. Obviously, we'll see how it turns out, but I'd be happy to wager a low stakes bet that half a year after launch of Wikifunctions we will have more than 200 active contributors. --DVrandecic (WMF) (talk) 00:50, 20 January 2023 (UTC)

Flawed calculation

This is great thinking of a kind that we should engage in much more often in the Wikimedia movement (to combat scope neglect among other things). For instance, I would also love to see a ballpark estimate of the number of people per month who read or look up at least one fact sourced from Wikidata (in whatever medium).

However, the calculation presented here has a fatal flaw, namely the assumption that unique devices can be used as a proxy for readers in this context. That is something that the team who created this metric has explicitly cautioned against. Sure, for a Fermi estimate, a factor of 3 (say) would not matter that much. But as soon as one compares estimated numbers as it is done here (by taking their ratio or difference), the results become rather meaningless.

To illustrate the problem, apply the same logic (steps 2, 4 and 5) to German instead of English:

2. German Wikipedia had 99 million unique devices last month (and 121 million in January 2021)

4. According to the data source you used, there are only 93 million German-language internet users.

5. So German Wikipedia reaches 106% of German-speaking Internet users, and in January 2021 it even reached 130% of them (divide (2) by (4)).

In fact, in Germany only a about a third of them use Wikipedia at least once per week, according to the well-respected ARD/ZDF-Onlinestudie.

Regards, HaeB (talk) 16:47, 30 December 2022 (UTC)

You are right that I was imprecise here. It would be more precise to replace "reader" with "unique devices". We explicate that in Footnote 4. I thought it would be easier to read the essay by using the term "reader" instead of "unique devices".

I do not agree that this necessarily invalidates the result, as long as we continue making the translation: i.e. instead of saying "700 million new Wikipedia readers" it should say "700 million new unique devices accessing Wikipedia" (neatly demonstrating why I chose the imprecision).

You are right that this adds another uncertainty to the estimate in some of the calculations (in particular, is the number of unique devices per person considerably different for the demographic that already uses Wikipedia compared to the demographic that is being added).

One major message was that the whole model should be represented in Wikifunctions and editable by the community - which then would allow for exactly this kind of collaborative improvements. I am very much looking forward to getting to that step!

Thanks for raising the point. --DVrandecic (WMF) (talk) 01:17, 20 January 2023 (UTC)