Free Lunch and the Square of Engagement 2009 January 2
One of the things I here often in the debate about Free/Open Source Software (FOSS) is that “there ain’t no such thing as a free lunch.” This phrase, at least according to Wikipedia article on Free Lunch that cites a 1874 New York Times article, originated based on a “custom peculiar to the Crescent City” (New Orleans), “in [which] every one of the drinking saloons which fill the city a meal of some sort is served free every day. The custom appears to have prevailed long before the war…. I am informed that there are thousands of men in this city who live entirely on meals obtained in this way.” The idea of the “free lunch” not really being free because the saloon owner made up for it in other ways. In other words the price of the food was factored into the drinks, thus the lunch is not really free.
In more recent times, it has been used to demonstrate opportunity costs. The thought is that even if you get something for free, in order to make use of it, you either can’t do something else, or you have to invest time to make use of what is giving to you for free. There are also other uses of this phrase in science, economics, and other areas.
So how does the Free Lunch apply to FOSS and is it really relevant? As Wikipedia points out, “TANSTAAFL (There Ain’t No Such Thing As A Free Lunch) is sometimes used as a response to claims of the virtues of Free Software. Often, proponents of Free Software, like Richard Stallman will say Free Software is Free as in Freedom, not Free as in Beer to respond to this criticism. While this is undoubtedly true my guess is that the people laying this criticism on Free Software are not going to be swayed so easily by Stallman’s virtues. Let us investigate this to determine is there really is some software that is free as in beer….
First, lets look at the claim that there is no such thing as a free lunch. Is this really true? I know I have had free lunches that didn’t cost me anything, so it seems a bit specious to me. In Dr. Friedman’s book, There’s No Such Thing As a Free Lunch, he says that “. . . every bit of pork, every subsidy, every inflation, every politically-imposed inefficiency, is paid for, and paid for in full — by somebody” (quoted from the book jacket by S. Leon Felkins). [1] Now Dr. Friedman was mostly talking about “free” handouts from the government, and as Felkins says “I suppose it is a true statement.” However, it is only true because of the last part, “by somebody.” In other words, they might be paid by somebody, but not by the person who receives the benefit. Thus, they recipient truly does receive a free lunch. If someone wants to pay for me to have a free lunch, I have no problem with that. And I’m not going to say it wasn’t truly free, because it was truly free to me.
Let’s look at a concrete example of this. One day I was driving across the Delaware Memorial Bridge into Delaware from NJ. When I made it to the toll booth, I found out the person in front of me paid my toll. I have no idea why. I have never seen the person before and as far as I know since, they just did it. Maybe I let them merge in or something earlier, I don’t know, but I didn’t complain. This was definitely the toll booth equivalent to a free lunch.
Now lets move this into the realm of software. The free lunch debate around FOSS usually comes up in the form of, well you can get program X for free, but now someone on your staff will have to support it. What does this mean? Well, I’m sure each person making this argument has there own definition, but basically the argument is that “You either pay to acquire it or pay to deploy it or both.” (Matt Assay)
Does this argument that their is no free lunch with Free Software hold water? Well, I think it might depend on how you look at it, but in a lot of cases I say no. Lets take a look at two different Web server applications as an example. If I need a Web server I can install Microsoft’s Internet Information Services (IIS) or I could install Apache’s HTTP (Web) Server. One of them, Apache, is FOSS. Is it Free Lunch too? Well, the argument that it isn’t free lunch would say that you have to pay to acquire it or deploy it. Well, I don’t have to pay to acquire it, so lets look at deployment costs. Since I have worked with Apache before, there is not much time for me to deploy it, but obviously it takes *some* time. However, how is this different from IIS? Deploying IIS isn’t done magically even if I pay Microsoft for the right to use IIS. Not only would IIS require me to pay to acquire the software, it makes me pay to deploy and maintain it as well. There in lies the crux of the problem with the whole Free Lunch debate when it comes to Free Software. The people that bring up that there is no such thing as a free lunch disregard the amount of work it takes to deploy proprietary solutions. Often the proprietary solutions take the same, if not more, staff time to deploy and maintain when compared to FOSS.
This concept applies to many FOSS programs. Look at Firefox. Firefox seems like a free lunch to me. Yes, I have to install it, but once I do it pretty much maintains itself. Yea, I had to spend about 2 minutes downloading and installing it, but even with a free lunch you have to chew. So maybe nothing is free, but even that expensive lunch you paid for has other costs involved. I remember an example in the library world when a university signed on to be a development partner with some project a proprietary automation vendor was releasing. They put out job ads to hire three people to implement the product. Meanwhile, another university had less then three people developing a similar FOSS project with less than two full-time developers. Not only was the first University paying for the product, they were paying more in staff time then the other university just to implement it.
Maybe Free Software isn’t often the same as Free Lunch. However, Proprietary software isn’t free of other costs either. However, looking at many of the Free Software products I use, many of them I do consider Free Lunch. Yes, I might have to take out a steak knife and cut up that free rib-eye someone offered me, but I’d have to do the same thing if I paid for it at some fancy restaurant.
Randy Metcalfe, the eIFL-FOSS program manager, recently blogged about the “square of engagement” that libraries have with FOSS. Basically the idea is that the for each FOSS application a library uses, you can plot it on a square. Across the horizontal axis is how much use that program takes, and vertically is how much engagement the library has in the application. Engagement can range from following a discussion list about the program, helping write documentation, replying to questions on e-mail lists from other users, to being a full-fledge developer of the program. In my way of thinking, programs that are used extensively but require little or no engagement are the Free Lunch of Free Software.
Randy Metcafe's Square of Engagement
With the square of engagement in mind, I decided to take a look at five of the FOSS applications we have been working with at some level at Binghamton University Libraries. The ones I’ve chosen (although there are a number of others I could have looked at) are Firefox, Apache HTTP (Web) Server, Movable Type, LibX toolbar, and E-prints. Since I don’t have any matrix to determine use or engagement, I estimated where these would go on the square of engagement. As you can see, we use Firefox extensively, but really have little engagement. The same is true of Apache. These programs are Free Lunch.
Selected FOSS applications at Binghamton University Libraries plotted on the Square of Engagement
Movable Type and LibX we use a little less, but we have slightly more engagement. E-prints, since it is related to a new product that we are just getting off the ground has more engagement and less use. I think looking at FOSS this way is really useful. By exploring where different applications are on the square of engagement, you can look to see if your library is engaging in the FOSS programs it should. If we are using something a lot, it makes sense to look if we are properly engaged in the community. While the answer depends on each individual library, I think at this point with these products we are probably in a good place. For example, it really doesn’t make sense, even though we use them extensively, for th library to spend resource developing Apache or Firefox. While they are very important applications for us, they are not software applications that are directly related to our core mission or that we necessarily have core competencies in. Also, there are a lot of talented people already working on these projects.
EPrints we have a high level of engagement with, but little use. This might be troubling considering the low use but for the stage of the project. As we learn and implement the software, it seems obvious that we will have more engagement than use. As the project involving EPrints at Binghamton matures, we will see more use. It remains to be seen how engaged we will become, but I do not expect a drastic reduction in engagement and we may become even more engaged. EPrints software is core to our role as an academic library so it makes sense that we will stay engaged in the product and not just become predominately users with little engagement as we are with Apache and Firefox.
I’m a big fan of open source, and prefer to use open source software at work when I have a choice. (I also prefer to write it).
And certainly there is a lot of misinformation going around about open source, some spread by proprietary software vendors.
One big one is ignoring the fact that for SOME open source software, you CAN purchase a support contract. The proprietary vendors like to pretend that open source and support contract are mutually exclusive.
Of course, that would be one example where use of FLOSS software would NOT be free of cost, when you buy a support contract.
I think it’s important that our administrators realize that FLOSS software doesn’t _automatically_ mean “and you get this without any money or staff cost to yourself at all.” The “cost of ownership” can vary for open source, sometimes it is indeed very very cheap–sometimes it is not, even if it’s FLOSS.
Our administrators are, let’s be frank, and easily confused bunch. Also a very risk averse bunch. If the promoters of open source inadvertantly give them the idea that “open source is free (as in $)”, then they’re going to think they never need to put any staffing resources into open source. Which is often not true. And the first time this strategy fails, they’re going to be turned off open source forever. It’s important to give them an accurate picture of what’s going on.
And usually, in my opinion, open source is NOT free as in $. It is sometimes, often even, _cheaper_ than a proprietary alternative. Certainly we must not forget that proprietary solutions are often a huge sink of staff time too. But individual open source products will vary in their total cost of ownership, and will _seldom_ be “free”, even if they are much cheaper than a proprietary alternative.
I’m not sure if you agree or disagree, Ed?
Hi Jonathan,
You bring up some pretty good points. I think I pretty much agree with what you say. I am not sure about “usually free as in $” as that really depends on what scale project you are talking about. I would say most of the Open Source solutions I use at work are approaching zero costs. I say that in good faith because the alternative (e.g. doing mass edits by hand instead of using sed) would have cost us much more time. The question becomes on what scale. Surely, implementing an Open Source ILS would not be cost free. It may (or may not) save us money over our current offering from Ex Libris, but regardless it would not be free. Whereas, I do believe that something like Firefox is free as in $. It all depends.
I am also not sure I’d agree that “Our administrators are, let’s be frank, and easily confused bunch” although I think this belies an interesting and important question. Certainly there is some truth to what is behind the statement and some administrators are more easily confused then others, however what is causing the confusion in our administrators about Open Source? These causes of this confusion is probably worth looking into at some point, but the fact is you are right that administrators don’t necessarily know what it means to use Open Source. Part of it is that most administrators were around before the FOSS movement took hold and they have long-standing relationship with library vendors who are trying to market their products who supply them with reasons not to go Open Source, but of course, most good marketers won’t tell you all sides of the story. Also, some of them may have lived through earlier “home-grown” solutions of the past that, in many cases, did not necessarily pan out all that well. I also think in many cases they do not understand how much staff time is put into proprietary solutions. I think it is the job of people like us to give them good recommendations. I agree with you that if all we say to administration is “the idea that “open source is free (as in $)”, then they’re going to think they never need to put any staffing resources into open source.” We really need to let administration (and other librarians) know how much time different projects require and what the pros and cons of a particular application is. I also think we need to discuss how the Open Source model can benefit libraries in many cases even if it is not the cheapest solution.
While my philosophical side wishes I could say that we would only use Open Source, it isn’t practical. The reasons vary greatly. Sometimes their is no available Open Source alternative. Some times the switching costs are too high. Sometimes there are campus or consortia politics. And so on.
Really, each project should be evaluated on its own and Open Source shouldn’t automatically be selected because it is “free” or be rejected because it is not proprietary or is “unsupported”. As you point out, often there are really good companies offering support for Open Source (Liblime and Equinox in the library world for example). The amount of time that needs to be invested, licensing costs, training, features, all need to be evaluated with any major software decision. We try to do that here at Binghamton. Yes, we are going with the Open Source EPrints for one project, but we also are quite pleased with some of the proprietary products we make extensive use of.
Some will say that you pay with your time or with your mental energy. I use Firefox, Apache and other FOSS but I am minimally engaged in their development or promotion. I tell my friends and family to use Firefox because I think it has great utility. The cost of installation ( though minimal ) was absorbed by me. All 2 minutes of my life. And the payback is that I no longer have to spend 2 hours cleaning the spyware crap off my brother-in-laws computer. So, no there was no “free lucnh” with Firefox. I got paid to eat!