I think the big losses are from the all of the wasted resources being put into making trades faster and faster. It seems to me that if merely swapping one set of machines out from a data center near the exchange and moving a few miles away is going to drastically change how that it performs, something is wrong.
From my personal experience, a brilliant friend of mine left his job at some Silicon Valley startup where he was definitely creating value for society to work at an HFT shop. After a few months of working there, he calls me and we talk about some C++/algorithms stuff (I think he is much, much smarter than me, I just happened to have written a lot more C++ than he has) about how he needs to make his code run microseconds faster so that it can make more trades.
I am all in favor of people who change the world for the better being rewarded immensely. I have no love for the iphone, but since many, many people do, I am sure Steve Jobs is worth his billions. But for people getting rich doing HFT, the only return for society I see are some nebulous claims about liquidity. Can the societal gains be quantified in some manner? If every transaction on the exchange had an added latency of a microsecond, how would things change? What if it was a millisecond? What about a second? How much worse off would the world be?
I worked in Silicon Valley for 12 years. Many of the projects I worked on had negative financial value - they never made any money. So essentially I was just wasting some rich person's money. The projects that were successful in terms of users and "eyeballs" often had negative social value. The whole point was to induce users to spend more and more time on our sites. Finally, some of the things going on right now in startup-land are quite frankly, creepy. The last project I did in Silicon Valley involved working with all these vendors who essentially are building up dossiers of every human who uses the internet and shopping it around as "market data." When I realized I was basically contributing to a surveillance state, I quit and moved to Manhattan, where at least I'm only fucking around with rich people's money, instead of fucking around with people's lives.
But you could also see it as "pushing the limit", similar to people trying to see who runs 100m the fastest, or Formula 1, or Kasparov-beating chess machines. Of course, in and of themselves, these look like silly activities and a total waste of dollars and energy given the poverty in the world. It strikes me, though, that these boundary pushing activities are somehow useful to the rest of us, in some weird and non-obvious way sometimes. Also, given that technologies have a habit of moving down the exclusivity scale, perhaps this persistent boundary pushing in HFT is contributing to general liquidity (speed of execution, lack of price impact, etc.) in a trickle-down way
Liquidity also isn't a nebulous concept. Look at over-the-counter fixed income trading markets, where prices are governed by ratings agencies and salesmanship as much as real price discovery, for a telling case study.
One way to look at the current situation --- probably idealistically --- is that HFT are to broker/dealers what ITA and Kayak are to travel agencies. That what liquidity is supposed to mean, right?
Shrinking times also have the advantage of improving price precision, when your talking about price moves over a very small period of time we can often be talking about very small price moves (say one-thousandth of a cent) which helps everyone. When before something may have been worth 213.314 you may have only been able to sell at 213.31, which for large deals can have a significant impact in high-volume markets, but with higher precision prices are more accurate.
But yeh basically the counter argument is liquidity. There's nothing stopping an exchange restricting speed of trades, the reason most of the major ones don't is companies don't want to be listed on a non-liquid exchange.
If HFT damaged the companies being traded than the companies could move to exchanges that didn't allow it.