Paying rewards to independent security researchers for finding software problems is a vastly better investment than hiring employees to do the same work, according to researchers from the University of California Berkeley.
Their study looked at vulnerability reward programs (VRPs) run by Google and Mozilla for the Chrome and Firefox web browsers.
Over the last three years, Google has paid $580,000 in rewards, and Mozilla has paid $570,000. In the course of those programs, hundreds of vulnerabilities have been fixed in the widely used products.
The programs are very cost effective. Since a North American developer's salary will cost a company about $100,000 with a 50 percent overhead, "we see that the cost of either of these VRPs is comparable to the cost of just one member of the browser security team," the researchers wrote.
Additionally, more eyes on the code meant the VRPs uncovered many more software flaws than just one hired developer could find.
The study provides a sound foundation for reward programs, which are not embraced by all vendors. Adobe Systems and Oracle do not pay for vulnerability information.
Microsoft has traditionally not paid bounties, but did implement a one-off program last month. Through July 26, Microsoft will pay up to $11,000 for bugs in its Internet Explorer 11 browser.
Bug bounties have other advantages, such as by reducing the number of vulnerabilities that are sold to malicious actors who would use the information for criminal activity. The programs also make it harder for hackers to find vulnerabilities, the researchers wrote.
But a key difference between Google's and Mozilla's programs may affect their effectiveness.
Mozilla pays a flat $3,000 reward for a vulnerability. Google pays on a sliding scale, which ranges from $500 to $10,000. Google judges vulnerabilities and exploits on factors such as difficulty and impact.
Google's average payout is just $1,000, but the chance of obtaining a much higher reward appears to provide an incentive for more people to participate in its program, the researchers wrote.
Google's program, while costing about the same as Mozilla's, "has identified more than three times as many bugs, is more popular and shows similar participation from repeat and first-time participants."
"This makes sense with an understanding of incentives in lotteries: the larger the potential prize amount, the more willing participants are to accept a lower expected return, which, for VRPs, means the program can expect more participants," according to the paper.
Also, browser penetration contests such as "Pwnium" run by Google with rewards up to $150,000 sparks more interest among researchers.
"We believe this sort of 'gamification' leads to a higher profile for the Chrome VRP, which may help encourage participation, particularly from researchers interested in wider recognition," the paper said.
"Accordingly, we recommend Mozilla change their reward structure to a tiered system like that of Chrome," it said.
The paper was authored by Matthew Finifter, Devdatta Akhawe and David Wagner.
Send news tips and comments to firstname.lastname@example.org. Follow me on Twitter: @jeremy_kirk