In my opinion and experience it's better to be the younger version who runs lots of experiments. While the older version may feel more wise, and may have taken

In my opinion and experience it's better to be the younger version who runs lots of experiments. While the older version may feel more wise, and may have taken some lessons from the past that helps him be aware of pitfalls in the future, there's always the danger of hindsight bias. The older version has ideas about why referrals didn't work, or the counter didn't work, but they might not be right (and I suspect they're not exactly right as Duolingo still contains different versions of affiliate referrals and counters). A better strategy is to think about it in terms of tradeoffs - if I build an affiliate program, what does that mean I won't build? I got the sense from the article that they explored one growth idea at a time, as opposed to evaluating a group of options from the beginning. All things being equal it's probably best to spend a bit of time thinking about several opportunities before going off and building the first one. Previous product experience won't guarantee you can call what will work for the future, but it can help you weight options vs one another more effectively. You're much more likely to have a nuanced view of what's possible or what might go wrong.