I was sad to hear of the passing of John Nash and his wife Alicia this weekend. May they rest in peace. As a game theorist I am familiar with his work and it just so happens that Nash Equilibriums have been in the center of what I’ve been working with lately. It’s an honor to be in a position to build on his ideas and hopefully pass his legacy on to a larger audience at the time of his death. I’ve done my best to pay him my respect by carefully studying his work over the past year, given Velocity starts this week.
The Prisoner’s Dilemma and the devOpsSec Dilemma, defined here as a lack of cooperation that stems from a lack of trust in an hyper competitive environment, have the same flaw: There’s not a game position that’s safe for everyone because of obliviousness or malicious intent. What I believe is emerging is a low trust cooperative game state. In a devOps world everyone means Everyone both internally and externally. This includes the unique identities of you, your co-workers, the company you all work for, the customers that keep you in business, and even the people who you perceive as your competition.
The oblivious trampling and opportunistic predation of other people and organizations are really complementary set operations that mitigate any risk associated with a perceived threat, which is what security is ultimately about. Staying in the lowest risk, highest paying game state is a decent definition of a secure bet for the short game but may not be a safe one over time. If the strategy is damaging others you will eventually compete yourself out of existence.
We have harassment problems in our industry. I want people to come to events and interact in an effort to network with folks because we grow faster when we share ideas. It would seem we have some of the wrong people showing up given the problems stemming from this type of interaction. If you’re passionate about technology and can respect other people and their skill sets you’re welcome in our community. If not, Keep Out. This applies to everyone, including pushy recruiters and vendors.
The problem with amensalistic behavior is you’re not aware that your offending someone when it happens. I’ve talked with a lot of people about this and there’s definitely room for a misunderstandings and they’re often cultural. When it turns into harassing behavior and you start looking like a parasite is when it’s characteristically repetitive. The key here is to pay attention to people. Watch for non-verbal queues when people are trying to politely excuse themselves. If you don’t catch the hints people will tell you they don’t have time or the energy to talk and you really need to take No for an answer.
I’d like to make it particularly clear that if you’re looking for a date you should go look somewhere else. Given the gender ratio issues at events your time would be best spent elsewhere. There are appropriate times and places for it but a professional environment is not one of them. Most conferences and events are in big cities. You can take your name tag off at any time and find plenty of trouble to get into outside of an event if that’s your thing. That means no one should have to ask you twice, Period. If someone has to tell you three times expect a less than friendly visit from event organizers. Let’s be clear that there’s also zero tolerance for unwanted physical contact in our community. No one should have to explain that to anyone and there is no good excuse.
With the exception of physical contact and other overtly offensive behavior, we also have to trust that people can make a mistake instead of automatically assuming the worst intentions. An intelligent agent will correct a poor behavior if they’re politely made aware of it. Again, if there’s a characteristic repetition in spite of someone asking them to stop their plausible deniability goes out the door with them. I noticed someone handing out business cards to a strip club in front of a conference recently. I told him that while it was mostly guys here it wasn’t the time and place to be handing those out and he’d have a problem on his hands if he didn’t leave. He apologized and left. Peer moderation can be effective and doesn’t have to hurt anyone. I feel confident this is a fair approach given the explanation of Blameless Problem Solving found on Etsy’s 2014 Progress Report:
“Making mistakes is an inevitable by-product of doing innovative work. Accidents can actually be valuable and rich sources of learning. We strive to create a blameless culture, in which it is safe to make mistakes and to speak up about them. This allows us to gain as much knowledge as possible from our experiences.”
We need to move out of no trust interference and do a better job of working together. This is a shift from no trust security towards high trust safety. This is also the split between devOps and security. DevOps believes in intelligent actors. Security assumes the worst of intentions. I covered this material at DevOpsDaysNYC. The video goes into more detail and has some ecological examples of the relationships on the chart, which I’ve added to since the talk.   The Prisoner’s dilemma is a Pareto inefficient Nash equilibrium because there’s no honor among thieves. Not defecting on the other prisoner leaves an opportunity for them to defect on you, walk out of jail, and leave you incarcerated for the longest period of time. Both prisoners defect and they both end up serving a longer sentence than if they had taken the risk of cooperating.
| ||Prisoner B stays silent (cooperates) ||Prisoner B betrays (defects) |
|Prisoner A stays silent (cooperates) ||Each serves 1 year ||Prisoner A: 3 years  |
Prisoner B: goes free
|Prisoner A betrays (defects) ||Prisoner A: goes free  |
Prisoner B: 3 years
|Each serves 2 years |
A significant change in a system replaces the original with a different, adapted system. A minimal number of local changes can be accommodated without inherently changing a system but enough local change in a short period of time is equivalent to a global shift. The devOpsSec Dilemma is the type of systemic failure that demands a systemic change. So I came up with a new game board.
Here we move from no trust competitive environment where everyone gets hurt towards a Pareto improved equilibrium. By attempting to move out of competition we risk being stepped on or preyed upon if our counterparts don’t make the shift to a higher trust relationship with us. The good news for our game board is that I don’t think anyone has to die or go to jail. While there is some risk here it’s most often a blow to one’s ego, which is a risk I can entertain for something I care about. The solution here is more diplomatic peer moderation and the ability to take constructive feedback even if it’s uncomfortable. We need to move out of safety and danger towards cooperative learning. When you move out of learning and into danger we end up hurt, but it isn’t always competitive.
Cooperation is at the heart of John Nash’s work. Trust is fundamental to our survival and is built by showing what you can contribute, not haggling and spit-balling in the bike shed. The more we trust the more we learn and the less we get hurt but it takes a combined effort. If your patterns as an individual or an organization don’t demonstrate these qualities over time then you’re done. As Deming put it, “Survival is not mandatory.”