It's meaningless to compare realized returns; it's only interesting to compare expected returns. Like if you say you're better than me at poker, it's pointless to play one round and compare our winnings, since randomness of a single round of poker far exceeds any difference in our skills.
Going with the poker example, your skill is higher than mine if your expected return per round is greater than mine.
One way to get a decent estimate is to repeat the experiment many times, while keeping external factors roughly similar. With poker, it's relatively easy: let's go to the same casino, spend a whole night playing at different tables (to avoid the luck of playing versus all noobs or versus all pros at a given table). Then compare our winnings. Better yet, repeat it over a few days, to get even more data points.
But in the case of a single large decision, such as Dell's investment, it's incredibly hard to estimate the expected returns. Maybe Dell's investment was an amazing decision, and he was all but certain to make $50B from it - and only due to an unusually unlucky random events, he only made ~$12B. Or maybe his investment was really bad, and he was likely to lose all his money, except through sheer dumb luck things actually worked out to a $12B profit.
If Dell went into the turnaround business, like Buffett, we could at least average out his performance over a dozen companies he tried it on. It would be still a very very rough estimate (e.g., due to survivor bias: Dell wouldn't turn around a dozen companies if he was losing a few billion per attempt: he would stop after 1 or 2). But at least it would be something.
As things stand, I doubt anyone can make a convincing case about whether Dell's investment was a good or bad idea.
By the way, even for S&P 500, it's not easy to estimate expected returns. But perhaps, over the 5 year period, they are around 30-60% -- depending on your opinion about various economic and financial models and assumptions. So I guess they were a bit better than expected over the period in question.
Going with the poker example, your skill is higher than mine if your expected return per round is greater than mine.
One way to get a decent estimate is to repeat the experiment many times, while keeping external factors roughly similar. With poker, it's relatively easy: let's go to the same casino, spend a whole night playing at different tables (to avoid the luck of playing versus all noobs or versus all pros at a given table). Then compare our winnings. Better yet, repeat it over a few days, to get even more data points.
But in the case of a single large decision, such as Dell's investment, it's incredibly hard to estimate the expected returns. Maybe Dell's investment was an amazing decision, and he was all but certain to make $50B from it - and only due to an unusually unlucky random events, he only made ~$12B. Or maybe his investment was really bad, and he was likely to lose all his money, except through sheer dumb luck things actually worked out to a $12B profit.
If Dell went into the turnaround business, like Buffett, we could at least average out his performance over a dozen companies he tried it on. It would be still a very very rough estimate (e.g., due to survivor bias: Dell wouldn't turn around a dozen companies if he was losing a few billion per attempt: he would stop after 1 or 2). But at least it would be something.
As things stand, I doubt anyone can make a convincing case about whether Dell's investment was a good or bad idea.
By the way, even for S&P 500, it's not easy to estimate expected returns. But perhaps, over the 5 year period, they are around 30-60% -- depending on your opinion about various economic and financial models and assumptions. So I guess they were a bit better than expected over the period in question.