This article proposes a beneficial impact of a deceptive practice in high-frequency trading. The deceptive practice is called spoofing:

In financial markets, to “spoof” means to make a bid or offer for a security or commodity with the intent of cancelling the order before it is executed. It is designed to create a false sense of investor demand in the market, thereby changing the behavior of other traders and allowing the spoofer to profit from these changes. It is illegal under federal statute.

The practice is illegal because is a deliberate deception on the market. The article states that there has been no case made that the practice is actually harmful or more harmful than beneficial. The fact that it is deception is sufficient for the law. The article makes the case that spoofing may be beneficial to the market by providing a needed countermeasure to the high-frequency trading tactic of front-running:

The most notable, and perhaps the most harmful, of these is what market players loosely call “front-running.” A front-runner profits by gleaning the intentions of legitimate market participants and jumping in front of their orders, thereby causing the original traders to buy or sell at a less favorable price.

The target of the front-runner attacker can use spoofing as a countermeasure to make the front-runner lose money on his tactic.

This is a real-world example of the problem I discussed earlier about how people can manipulate data to get foreign analytics to behave to ones benefit.  The above article mentions a recent prosecution of someone who actively employed spoofing in his market practice.

Spoofing is deliberately injected data for the purpose of it being observed by the foreign sensors and then incorporated in the analytics for automated decision making. It is a form of hacking that uses data instead of malware code to make the foreign system to behave in a way intended by the attacker. In this case the data is the spoofed order. The attack is especially effective in high-frequency trading where the time-window for any financial advantage of the front-runner is too fast for humans to notice.

The high-speed nature of this cat-and-mouse game illustrates my earlier post about data analytics being like an operating system.

In this example, the front-running algorithm is an example of a natural strategy to take advantage of big data analytics, in this case to recognize opportunities faster than humans. There is nothing deceptive about front-running but there is a case to be made that it is unfair to human traders and especially traders involving large transactions. The spoof algorithm is an example of a deliberate injection of data to disrupt the foreign front-runner algorithm either by misleading it into cooperating with the intended transaction or by imposing a high loss on the algorithm to discourage it from targeting the same trader.

This example also illustrates governance in the form of outlawing and prosecuting the deceptive practice of spoofing.  It will still happen, but at least there is the risk that the spoofer will get caught and face a high penalty.

This also illustrates the difficulty of determining the good from the bad for the purposes of data governance.  From a purely ethical point of view spoofing is dishonest deception while front-running is honest taking advantage of a structural timing advantage.   On the other hand, government regulators of the financial markets are more interested in protecting the interests of traders, especially smaller traders from being robbed of their potential earnings.  From their perspective, the spoofing, when used specifically as a counter-measure to front-running may be something to encourage rather than outlaw.

Governance involves regulation of some sort, but that regulation would have to be as high frequency as the analytic tools in order to separate the good forms of spoofing from the bad forms.  Regulation is not that responsive so the governance is the sluggish and potentially ultimately harmful categorical outlawing of spoofing.

When it comes to high velocity data analytics, data governance is very difficult to get right.