This is really a pretty interesting, thought-provoking article. I think the basic point is that people (players [good players, not the clueless masses] and casinos) focus on EV and ignore volatility, but that it's the volatility that really affects longevity in the case of slots.
This really makes perfect sense. From a mathematical standpoint (sorry, folks), EV is the mean and volatility is the variance (which is the square of the standard deviation, if you're really into such things). The mean is approached only in the long run (for example, you'll keep 99.54% of your money playing 9/6 JoB for a very long time), but in the short run, volatility is King. Now, the "long run" is also determined by the variance; the higher the variance, the further away the "long run" is. The variance of many casino games is low: it's 1.3 for blackjack, 1.0 for roulette, 2.7 for three card poker, etc. It's 19.5 for 9/6 JoB, up to 42.0 for DDB. But for slots, it's much higher, in the 50-120 range, probably even higher for the most volatile machines like Megabucks. Think about it, how long does that $20 last on Megabucks? And how long would you need to play to get that jackpot at least once, to get you to the long-term EV?
Now, the article talks about a simulation where a 97% EV slot and an 88% EV slot yielded about the same longevity, probably because the variance was about matched. I think the take-home lesson is that in high-variance games (like slots, probably also keno and volatile VP games like DDB), you won't feel the benefits of a better EC for a VERY long time, like over the span of a lifetime. On the other hand, you'll have a more predictable ride with a lower volatility game (like a lot of the better VP, and blackjack).
Thanks to Rec, by the way, for pointing a lot of this out over the years.
