www.lesswrong.com/posts/k6nncsXAZD3bK6fe3/invisible-choices-made-by-default

The post was published on October 20, 2019. Historical Google Play tracking for the Android apps contradicts both download figures:

• Duolingo had already reached 10 million installs by December 29, 2013, 50 million by December 21, 2015, and 100 million by August 13, 2017. So by October 2019 it could not have had only 8 million Play Store downloads.
• AnkiDroid had already reached 5 million installs by September 1, 2019 according to the same Google Play history. Separately, AnkiDroid maintainers wrote in June 2019 that more than 1 million users have installed AnkiDroid, and in January 2019 that it had 1 million active Android users. So it could not have had only 40 thousand downloads/users on Android at the time of this post.

Because both inputs are wrong, the conclusion that there were "200 Duolingo users for every Anki user" is also wrong. Even if you use only Android install counts, the ratio was nowhere near 200:1; using the available October 2019-era thresholds, it was on the order of 100M to 5M, i.e. about 20:1 on that metric.

One additional caveat: Play Store installs/downloads are not the same thing as total users. But even on the post's own chosen metric, the stated numbers are incorrect by roughly an order of magnitude.