The explaination could simply be that mainly only smart money short stocks, and they tend to be pretty good at figuring out issues in companies that are not yet fully priced in.
If I were to run a short ETF, I would add filters such as market cap, short interest and day to covers to avoid liquidity issues and squeezes, and/or play it through options structures. Like, it would have been stupid to be short GME when it had 300% of float as short interest. But it is probably fine in a lot of case with a sufficiently low day-to-covers and short interest. A more active strategy would be to monitor the Fee Rate which gives a good indication of the stock supply and is more "real time" than short interest data which often lags a few days/weeks.
So if you pay 3% borrow costs you do worse than putting your money in Treasuries? 3% seems conservative for the most shorted stocks, I checked a few on [0] - TSLA 0.25%, WE 4%, AMC 143%. Uncorrelated returns are great but this is not the most compelling sales pitch for a short fund. Better to use this information to underweight those stocks in your long portfolio.
Treasuries are quite correlated with the stock market as both depend on liquidity, so it's not exactly a hedge, although they tend to do better during recessions, they will go down during unexpected hikes.
The best way to run a short ETF (on something where no futures market exists) is to also manage the long version of the fund and cap interest in the short fund to be less than or equal to the long. If it's a serious investment it's likely the long fund naturally has more interest anyway.