Algorithmic trading is first provided enhanced access by the market operator. Econometrically this provides assymetric participation and information inefficiency. Symmetry and efficiency are vital aspects of any market.
Secondly, Joe touches on the lack of obligation to continue participation in providing liquidity. Sharp falls occur when the vaccum is created in the buy side by the lack of buyers in the presence of sellers. This is an enduring principal that existed since markets were invented. But algo's have no responsibility to continue to trade like anyone else. So the liquidity they provide does not solve anything as algo's circuit break themselves in the absence of liquidity, and realise the opportunity when it happens in milliseconds - yet the market order books still do nothing to prevent prices crashing.
This could all go away if preferencial access and financial reward (from the market provider) was not given to algo traders. Also, a minimum order size and lifetime as well as universal access charges (not rebated) for all participants. Finally, add bid/offer matching within a fixed percentage of the last transaction using minimum order sizes and at least there is continuity in price flow that spans several seconds, and not milliseconds.
As for the role of markets to provide 'price discovery' - that's laughable in itself, as discovery is being destroyed by algo trading.
If you can't join them, try and beat them by other means - using better technical analysis (as this is not fundamental information). Buy and hold is dead - with more liquidity provided by human participation (by buying and selling) this would not be so prevalent.
The volume oftrading provided by algo trading can be fixed if someone was prepared to do something about it.
