One important thing to note is that although the far right end appears to narrow, that is not because returns become more certain over very long time ranges, bu

One important thing to note is that although the far right end appears to narrow, that is not because returns become more certain over very long time ranges, but rather simply due to lack of samples over those very long periods. What's more indicative of reality is the relatively constant vertical distribution, in a log scale, across any time range up to about 90 years, when the lack of data starts to play a significant role. So, in relative terms, uncertainty remains fairly constant over time. In absolute terms it actually increases. In a sense risk still does decrease over long time horizons, as the chance of earning over any given amount does increase with time; contrary to common belief though, the distribution of possible returns (another possible definition of risk) actually increases with time - or remains roughly constant on a log scale, as mentioned. (You can find a great many opinions surrounding this by googling "time diversification".)