The risk-free (i.e. government-guaranteed) inflation-adjusted 30-year interest rate in the U.S. is about 1% at the moment. On the one hand, that a seems depressingly low, and compared to historical rates it is. On the other hand, periods of low long-term real interest rates tend to be highly correlated with periods social and political stability, so perhaps today's low interest rates are a price worth paying.
If you are willing to accept a reasonable, but non-trivial, amount of risk, you could invest the stock market. A 3.5% inflation-adjusted rate of return is actually a very solid guess about future long-term stock market returns. Of course, there is definitely a risk that your returns will end up lower - that's why the stock market is a higher-risk, higher reward investment.
Here's a useful rule of thumb for estimating inflation-adjusted stock market returns in developed nations over long periods time (at least 20 years): Rate of return = Real economic growth rate + Divided rate - Expense ratio - Dilution rate. The "dilution rate" is the rate at which your shares of stock are diluted by companies issuing new shares of stock, and the the "expense ratio" is the proportion of your assets that are consumed by investment costs, usually in the form of transaction and recordkeeping costs incurred by your mutual fund.
The dividend rate of the domestic stock market is currently about 2% per year, and the market's dilution rate seems to be around .5% per year. Assuming that you invest via a low-cost indexed mutual fund from a company like Vanguard or Fidelity, your expense ratio should be a rounding error - less than .2% per year. The real economic growth rate is the only unknown, but for the past two centuries has remained surprisingly close to 2% per year in industrialized nations. Again, keep in mind that we're talking long term trends; obviously growth rates were much lower during the recent recession and much higher during the .com boom.
Putting that all together: 2%+2%-.5%-.2% = 3.3% estimated rate of return, which is almost exactly the rate cited by the grandparent. Keep in mind though, that this is a very long term estimate; the stock market might go up or down 50% in a single year. Also keep in mind that there's a good chance that this estimate will be wrong - after all, the reason that the expected return is relatively high is that there's a reasonable but non-trivial risk that your rate of return will be much lower than expected, even in the long run.
Anyone who wants to become a finance nerd would do well to read William Bernstein's book The Intelligent Asset Allocator. The book explains the rationale behind this rule of thumb, and everything else you might ever want to know about estimating financial returns.