We could assume: popularity = corporate_hype + 0.3 * quality + random()
...and despite the correlation, things that both very good and very popular are really rare compared to very-good _OR_ very-popular
We could assume: popularity = corporate_hype + 0.3 * quality + random()
...and despite the correlation, things that both very good and very popular are really rare compared to very-good _OR_ very-popular