Predictably Irrational

I’ve been working on OpenNMS for over seven years now. Before that I spent over ten years with proprietary tools, including software from most of the big players like IBM’s Tivoli and HP’s OpenView.

When I first got involved with open source management software, I thought pretty much what everyone thinks: cheap OpenView. What I’ve learned in all those years is that free and open software is so much more than just “cheaper”. In fact, trying to sell open source software as a cheaper version of a proprietary product is the wrong way to go. It puts the buyer into the position of comparing the two products as if they were the same, and thus they only focus on what they are familiar with, the features of the proprietary product, and not the real benefits of the differences.

Free and open software does not mean cheaper licensing fees. It means freedom from licensing fees altogether. It moves the power to direct how the product is used and developed from the vendor to the consumer. This results in both the removal of vendor lock-in as well as the ability for the solution to grow with the business with out any “management tax” that per-node pricing models impose.

Unfortunately, the open core crowd, by using the term “open source” to describe their proprietary software, muddies the waters quite a bit. By positioning themselves as the “cheap OpenView” they end up competing in terms of the proprietary software, as in “do you have this feature” and not with “here’s how to manage your network better”.

I’m often surprised by the success of open core software, especially in the US. To me it is a bit irrational. Why give up per-node priced proprietary software with the attendant vendor lock-in for cheaper per-node priced proprietary software with vendor lock-in, even if there is an open component?

I’ve attempted to understand this with my Free Food analogy. Suggesting an open source software solution is like offering someone free food. If someone offers you free food out of context, you immediately think “is it safe to eat?” This is followed quickly by “what’s the catch?” In the case of open core, the decision makers, once they find out that, yes, there are software licensing fees involved, relax. They understand licensing fees and they are comfortable making a decision within the framework of “cheap OpenView”.

I was talking about this with Alex Hudson in London, and he turned me on to a book by Dan Ariely called Predictably Irrational. It’s a great read, and it demonstrates, complete with experiments, how we moist robots often act irrationally when making choices, but in a predictable fashion. Given a particular situation, one can reliably predict which irrational decision will be chosen.

As a businessman I need to be able to position the strengths of OpenNMS against proprietary products on a basis other than “cheap OpenView”. I need to be able to describe the power of the community, the flexibility in getting exactly the features you need in weeks instead of years, and the wonderful things that can happen when you put a tool like OpenNMS in the hands of a capable administrator. I need to make sure that, when presenting OpenNMS as a solution, that the decision to choose it is rational.

The first example in the book involves a magazine subscription. The buyer is presented with three options:

  • On-line content only: $59
  • Print content only: $125
  • Print and On-line content: $125

Notice that the Print and Print+On-line options are priced the same, so a rational person would never choose Print only (although I am certain a few Luddites would do it if the sample was large enough). However, Ariely found out through experiment that many more people would choose the $125 option over the $59 option when all three options were presented than if it was removed. The addition of this “decoy” option caused people to value the $125 option more, even though there was no rational difference between the two offers.

I can see how this would work. By seeing the three options my thinking might run something like this:

The Print option is $125 and the On-line option is $59. The Combined option is $125, which represents a savings of $59 since On-line is included free. If I subtract the $59 from $125 I come up with $66, so I’m actually getting the Print version for nearly half price!

Of course, this is complete bollocks. If my main source of accessing this information is on-line, then the Combined option actually costs me more than twice as much, but Ariely’s experiments showed that this is how many people make decisions.

So, what does this have to do with open source software?

In other sections of the book, Ariely explores predictable irrationality in a number of different contexts. While all of them have some bearing on selling in general, several really struck me as relevant to open source: “The Cost of Zero Cost”, “The Cost of Social Norms” and to a lesser extent “The Effect of Expectations” and “The Power of Price”.

What Ariely found out in his research was that the term FREE! has a profound affect on behavior. In one experiment they offered a Hershey’s kiss chocolate for 1 cent and a Lindt chocolate truffle for 15 cents. Since the truffle costs about 30 cents, it was the better deal, and people realized it.

However, when he lowered the price of each by one cent, making the kiss free, the situation was reversed. Many more people chose the kiss, even though the economic value was the same in both situations.

Being a geek, I immediately thought that part of the problem could have been the purchase process itself. I may not want to stop and purchase a candy but I might pick one up for free in passing. I could rationalize that my time has a certain value. One thing I love about this book is that Ariely takes this into account (he’s either just a good experimental behaviorist or a geek – maybe both). When the experiment was repeated as part of a cafeteria check out line (where the consumer was already in the process of purchasing so the choice of whether or not to buy candy added almost zero time to the transaction) the results were the same.

This would seem to be a good thing for open source, and in many cases it is. A harried sysadmin with little time or patience for the purchase process might grab MySQL or Apache, even on Windows, just because it is free.

The problems arise when you bring the idea of commercial open source into the mix. By commercial open source I don’t mean open core but commercializing an open source project in a manner that doesn’t involve proprietary software licenses, such as services and support. When you add in a cost, any cost, it seems to bring the discussion back into comparing software products and not solutions. Thus a product like OpenNMS now gets put into the same feature for feature comparison of a commercial software product like OpenView, even though the value offered is totally different.

When this happens a number of things take place. In one sense brand awareness and expectations come in to play. In the Expectations section of the book a particular food dish is described as a “mélange of the freshest roma cherry tomatoes and crisp field greens, paired with a warm circle of chèvre in a fruity raspberry vinaigrette”.

Sounds a lot better than “goat cheese salad”, doesn’t it? One would expect it to taste better than a plain ol’ goat cheese salad, too.

In the “Power of Price” chapter Ariely shows studies where patients experienced a more rapid and complete loss of symptoms when they believed they were taking a more expensive medication, even if the two choices offered were the same drug, or even when the more expensive medication was a placebo. Expectations and price play a big role in the buying decisions we make.

This affects OpenNMS, especially in the high end markets such as carriers. When a company is used to spending tens of millions of dollars per year on management software, it is inconceivable to them that they could get the same thing for free. Again the problem comes from comparing the price of OpenNMS software against, say, IBM software and not the solutions themselves, which are a combination of software and services. No one at the carrier level buys software without a huge services component, but in trying to position the services that The OpenNMS Group supplies against those from, say, IBM Global Services, it seems that it it often shifts back to the software.

I had to laugh at the recent row between Groundwork and HP over the practice of Groundwork comparing the price of its solution against OpenView. Groundwork is an open core software company that cobbles together a number of rather good open source tools. Unfortunately, in the accounts where we have replaced Groundwork with OpenNMS our experience with that product shows that it is nowhere near as well integrated as OpenView’s various parts (and that’s saying something) and it is definitely not worth the rather large price they place on it – OpenView or no OpenView.

Seeing Groundwork trying to compare itself to HP reminds me a lot of Hyundai’s latest campaign to compare its Genesis car against brands like BMW and Lexus. Due to a bad car accident I was in many years ago that was caused by a Hyundai driver, I’ve never liked the brand. Recently I rented a car and was given a Sonata, and I had to admit it wasn’t a bad car. It wasn’t a car I would buy, but for an economy sedan it had a lot of high end features.

The problem is that you can’t take a car that was originally designed to compete on price and suddenly move it into the luxury market. In people’s minds the expectation has been set that Hyundai is a cheap car. Not necessarily a bad car, but definitely not a luxury vehicle. The price and value of a Hyundai have been anchored in peoples minds (see the chapter “The Fallacy of Supply and Demand” in the book to learn about anchors) and once anchored it is hard to change. That’s why there is a Lexus for Toyota and Acura for Honda – it was necessary for those companies to break the existing preconceptions.

Open source software has almost always been marketed on price, and when that happens it will always be hard to break into the markets controlled by expensive commercial software. It is very important for the industry to focus on the non-software benefits – freedom, adaptability, ease of ownership and the benefits of the community – instead of trying to just be a less expensive version of an established product. Part of that is to be honest about the differences between open source and open core. Bringing closed software licenses into the model will just cause confusion.

I’ve often said that the main strength of The OpenNMS Project is its community. By being totally free software, and sharply separating the commercial services business from the project, we have managed to gather a small but dedicated core of individuals who are emotionally invested in it.

Ariely addresses this community in the chapter on social norms. He divides social interaction into two types: those governed by social norms and those government by the market. Severe problems can arise if one tries to mix the two.

His first example describes a Thanksgiving dinner. In the US the Thanksgiving holiday falls on the fourth Thursday of November, and it is usually celebrated by family gatherings and a large meal, quite often a feast. This feast is often prepared by a matriarch in the family (in my case it is my mother) He imagines what would happen at the end of the meal someone got up and offered to pay this person. I know in the case of my family my mother would be mortified. She cooks (for days) as a labor of love, and to try to turn that effort from a gift into work (which is what happens when you place a monetary value on it) is highly offensive.

Quite often businesses try to form a more social relationship with their clients due to the vast benefits it brings (I probably could not afford my mother’s Thanksgiving meal if I had to pay for it) but this can backfire. In the book there is an example of a day care center that had a problem with parents picking up their children late. It was a small business and the owners and the clients had a social relationship, so when the parents were late they felt bad about it and tried to be better in the future.

However, the owners decided to impose a fine if the parents were late in an effort to reduce the tardiness even more. The moment the fine was brought in to play, the dynamic changed from social norms to market norms. Now that parents were paying for being late, they decided to judge being tardy not in terms of inconveniencing the owners but in terms of the cost versus whatever was causing them to be late. In other words, they had a market way of determining how being late affected them without regard to how it affected the owners of the day care center.

There’s more. Since the fine actually increased the number of late pickups, it was removed. But the parents’ behavior didn’t change. Once the relationship moved from the social to one based on the market, it was very difficult to move it back.

This, more than anything, is why I am happy that we have been so vigilant about separating the community side of OpenNMS from the commercial one. The OpenNMS Project is a community owned and community run endeavor that is governed by social norms, and The OpenNMS Group is a business, run by market norms. We still strive for a social relationship with our paying clients, but the commercial side is a real, profitable business and not a charity. If we started to offer proprietary extensions to OpenNMS, I bet that part of our community would then start to view the project in terms of market norms and we would lose a good part of the benefit the community provides.

While I still have yet to digest everything I read, I did take away some key points:

  • Remember that they power of open source comes from the community, so make sure not to destroy it by introducing market norms into the relationship.
  • Competing on price turns a discussion of solutions based on open source into a discussion of software versus software. In many cases open source will lose.
  • The value of FREE! is strong, but the commercial open core companies use of the term “open source” dilutes that. We in the free and open software business need to be vigilant about educating the market on the difference.

I don’t recommend “business” books very often. In fact, I can’t remember recommending anything other than Crossing the Chasm, so take that into account when I say to run out and get Predictably Irrational: The Hidden Forces That Shape Our Decisions.


LinkedIn is the only social network I even remotely keep up with. Recently Matt Asay introduced me to TripIt.

TripIt adds Dopplr-like travel information to your LinkedIn account. While I liked Dopplr I never really bothered to keep up with it and it seemed to be a pain to have to re-link to all of my contacts.

After dealing with some “server too busy” issues, I managed to get next week’s trip to Chicago added. It doesn’t seem like many of my contacts are using TripIt, so I decided to pimp it here.

Probably my favorite aspect of my job is meeting people in the community, so if you see me in your neighborhood, be sure to let me know if you want to meet for a beer or three.

Work/Life Balance

I am back from my holiday in Fiji and New Zealand, but not fully back in spirit. That should change over the next week as I get back into the swing of things and it will take awhile to dig out from under the pile of e-mail and news.

We heard about the attacks in Mumbai, of course, but were shielded from most of the news coverage since we didn’t watch television or get on the Internet for much of the trip. My best wishes go out to those people affected, and I know that there are probably a couple of OpenNMS users in that group. In more bad news, a friend of mine was diagnosed with cancer in my absence, but I have no doubt that he’ll pull through. What doesn’t kill us makes us stronger.

It’s also December and the holiday season is upon us. Again, I’m not quite in the spirit as I just spent time in really warm, really sunny places, which is not the norm for December in North Carolina.

Believe it or not I didn’t think too much about OpenNMS during my vacation. When at home it occupies most of my waking thoughts and quite a few of my dreams, so I was surprised that I was able to unplug as much as I did. A good portion of that was due to the lack of Internet access. As much as I love the network and the opportunities it provides, it has really eroded the divisions between the work place and personal life, sometimes called the “work/life balance“. I’m pretty much in my office whether or not I’m actually there, at home or on the road. It takes getting out well out of my element for me to get away from it all, and this trip made that happen.

It was 44 hours from the moment we left Wellington until I got home, but the trip went as well as could have been hoped. I was surprised at the number of people on the domestic flights, but then I realized that it was the Thanksgiving holiday weekend which is known as the busiest travel weekend of the year. I have to agree with Louis CK that we don’t appreciate air travel as much as we should.

I wish I could do it more often, but a two week holiday once every other year is about all I can manage. I think the Europeans have it right when it comes to time away from work. I scrambled to see as much of New Zealand as I could in 10 days, and even limiting myself to the north island I missed a lot. During my stay I met a couple from Belgium and a man from Holland who were both spending five weeks in the country, which is more reasonable.

I did get to meet one OpenNMS user for lunch while in New Zealand. Daniel Lawson lives in Hamilton which was on our way out of Auckland.

It makes travel fun to know that almost anywhere I go I can run into someone from the community. Next stop: Chicago. Why do I always end up there in winter?