This might be an example of marketing to an encouraging interviewer who is already an admirer and receptive audience.
I have the performance record from his recent letter. For any holding periods less than 20 years, he has underperformed the S&P. In fact the record shows that he has underperformed in "every" holding period between 1 and 20 years which is by itself pretty remarkable. All the "outperformance" is if you started your holding between 20 and 24 years ago.
I think it may be a similar case with Pabrai.
To his credit he supplies the figures pretty transparently but the marketing and positioning tells a different story. This might explain why some these managers are guarded in sharing a lot of this information publicly.

Though I have no information about whether or not they are good investment managers, one metric for any manager that isn't very useful is the performance over the last 1/2/5/10/15 years. Which is of course what you most often see.

Why is it a terrible metric? because all those figures end today, so in one sense it's only one data point. Today might be a typical end date, making all the figures pretty meaningful. Or today might be an unusual end date, making all the figures wildly atypical to the upside or the downside.

Underperformance over short to medium to even medium-long periods might mean one of two things: the stuff in their wheelhouse is very cheap right now, or they are terrible managers. Those figures give you zero information about which situation it is. For example, virtually every outstanding value manager failed over almost any lookback you care to choose if you looked at all the return figures for all the various holding periods ending around the exact same time in 1999.

Check out the year by year returns. I like to look at a graph of the ratio of managed portfolio value to [appropriate] index portfolio value. A flat line is an index tracker. It is an uptrend with a dip at the end (everybody has a bad year) or is it a gradual downwards trend over many many years? Is it jagged, meaning a lot of big random positions?

Jim