Nobody will ever know the exact cause of this historic collapse (the largest intraday drop ever) followed by a recovery within a few minutes. It impacted several stock exchanges and most stocks, and the volume was heavy.
Is it a terrorist plot? No, terrorists don't have this level of sophistication. Even if the terrorist was a trader in a major US bank.
Is it manipulation by the government to extract hundreds of billions of dollars to finance health care, the Greeks and other problems? I don't believe in this conspiracy theory, because the government also lacks the required level of sophistication.
Wall Street crooks? I don't think so. Did money really change hands? Maybe for a few minutes, but it's not like 500 billion dollars were "stolen" from ordinary investors and permanently transferred to a few large accounts. It's likely that the money came back to those who lost it (see my section on "probable cause" to understand why it came back).
Trading error? No trading error could produce such a drop.
The stock market is driven by emotions. Behavioral analytic models explain swings and volatility. If you've been trading recently, chances are that (a week ago) you had many stop orders, and many short sell orders (more than ever before) ready for the largely expected moment when DOW goes back below 10,500. Also, you would have many limit (buy) orders when the DOW goes below 10,000 (it did on May 6). Chances are that most traders, small ordinary guys, big guys and automated trading bots and algorithms had these mechanisms in place based on human feelings associated with the recent spectacular recovery of the stock market over the last year.
With so many stop orders (and short sell orders), once we reach a tipping point (10,600) a cascading effect occurs naturally, resulting in the collapse. Think about the Titanic filling with water: it stayed afloat for 2 hours, then in a matter of minutes, it sank. The same thing occurred to DOW, with a similar time line. The difference is that DOW came back, because of the very numerous buy orders below 10,000.
In order to investigate the real cause of the collapse (whether it is natural or not), one has to look at the distribution of percentage drop for all major stocks. Since DOW lost 6 percents in a few minutes, if the cause is natural, you'd expect that on average, all stocks lost 6 percent. But there are thousands of stocks, so you would expect some (very few) will indeed go up, while some will lose more than 30% (Procter & Gamble).
How do we know if the 40% drop on Procter & Gamble is "natural" (being the most extreme drop for a large stock out of thousands spectacular drops), or not natural? To answer the question, one has to consider extreme value theory, or better, perform simulations. I have no doubt that with so many stocks and such a severe global drop, a few large stocks will experience a drop 2 or 3 times worse than average.