I think this app-button experiment should be paused, and postponed for a medium-to-long-term future, for a few reasons:
- The portal improvements are starting to be well-received, with the recent auto-adjusting language items, and the highlighting of smaller wikis. We should focus on building off this success, instead of risking a step backwards. There are much higher priority improvements for the portal page, including replicating these upgrades to the sister project portals (unless that is already done), and automatically updating the article counts, and translating the sister project descriptions.
- The readers probably arrived at the portals with the intent of searching for something, and the app-buttons are a distraction from that task.
- The app-buttons do not funnel readers towards the wikis, which is the primary goal of the portals.
- The apps do not currently support the sister projects, which frustrates editors and readers alike.
- The app-buttons are placed below/next to the sister project links, which is thus even more confusing.
- The app-buttons are harder to translate, given the graphical nature and custom fonts. Having the apps available to non-English speakers is probably a priority, and whilst they'll be used to recognizing the visual aspects of 'app download buttons', it would still be nice to translate this, too.
That said, and whilst I do understand some of the other reasons why some people are not fond of the apps in general, I also appreciate that a huge part of the world wants "an app" for a few reasons (hence the dozens of existing externally created apps in the mobile stores): I/they appreciate the speed of software load (versus my android browsers which both take an age); they also want the easy installation on the "home screen" that installing an app provides (the non-technically inclined are less likely to discover how to add https://en.wikipedia.org/wiki/Special:Search ). Plus the variety of additional features that apps (internal and external) can more easily experiment with. So I do grok the impetus to both create the apps, and to promote the existence of our apps (especially given that we are helpfully supplying some less personally intrusive apps than I would assume many of the externally developed apps provide). But I respectfully suggest that this is a poor moment in time to do the latter, at our Wikipedia portal page.
As for criteria as to when a good time would be, I would suggest at the earliest would be after some more milestones are reached, such as the tasks given in my first list item. A much longer milestone would be my 4th list item, but that's less pragmatic, and there must be balance.
If these points are not judged to be sufficient rationale to pause the experiment, then I urge the creation of some sort of "success criteria" before the experiment (currently missing from the docs page, though possibly it appears within a phab task?), so that we know how the metrics that are collected will be interpreted, and thus know how the success or failure of the experiment will be judged.