One current big controversy among economists is whether the HFTs themselves have an effect on volatility (there are studies claiming all of "increases", "decreases", and "have little effect", but the evidence for any of those conclusions is weak, and it depends on exactly how you define volatility). A liquidity provider who provided added liquidity on very short time scales, but at the expense of increased volatility at longer time scales, wouldn't necessarily be advantageous for longer-term market participants. That'd be essentially trading one kind of noise for another one--- less millisecond variation, and more occasional giant aberrations, if indeed things like the May '10 "flash crash" were caused by HFTs.