Ravi Mohan's Blog
Tuesday, September 13, 2005
A Nostalgia for Guilds And Other Dangerous Ideas
After some thought provoking conversations over email with Dave , Ade and Rajesh recently, I read Pete Mc Breen's book on Software Craftsmanship . The book's fundamental thesis is interesting. It says that the "software engineering" way of creating software is broken (I agree) and it posits a "craftsmanship" approach to software as the solution .(I disagree). Let me clarify that statement. I don't disagree with the idea that viewing software development as a craft rather than some kind of factory process will yield insights into how to progress along that craft. The basic problem with the book is that it proposes what is essentially a medieval socioeconomic structure as a solution to a 21st century problem ,and selectively focusses on the virtues of the old "guild" system where practitioners of a craft were classified into 3 categories - apprentices, journeymen and masters . Now let us examine how Mr McBreen proposes to solve the "software engineering problem" . he says, (all emphases mine) "Rather than attempt to industrialize and deskill software development, the software craftsmanship approach looks back into history to see how the problems of expensive labor and arcane skills were handled. " ( Chapter 5 ) While the book doesn't go into too much detail on which nation's history of what period is being "looked back at" , it is fairly obvious from the sorrounding context and talk of "Apprentice, Journeyman and Master" that he refers to the guild system that flourished in Medieval Europe. So Mr Mc Breen, on the one hand says that a methodology originating in the 1960s and continously evolving since then (software engineering) doesn't work for software development in the 21st century so we should go back to the 14th , for solutions ! Interesting ! There are two ways to use labels like "apprentice", "journeyman " and "master". One is that of a rough guide to where on a skill spectrum a practitioner of a craft falls.In any time period, in any craft or art, there will always be a fairly consistent ranking of people in terms of skill and it sometimes helps to mentally divide that spectrum into n categories and name those categories whatever you want.Yes, if n ==3 you can name the categories , "apprentice", "journeyman" and "master" .This is only a way of seeing the progress along a craft .Those labels have no real validity . The second way is to use such terms as rigid unchanging classifications of people(however derived) and attaching separate privileges to these labels . Thus you couldn't run a craft based business in medieval Europe unless you were a member of the appropriate guild and presumably of an appropriate "rank" to do so . Today anyone can start a business and thrive or crash based on the economic logic of the business. Which system would you prefer to live under ? the fact that Pete Mc Breen supports the second ,"rigid label", way is evident from a paragraph that states "A master craftsman may learn a new technology from an apprentice, but it does not mean that she is no longer a master. If anything, it reinforces her mastery because it illustrates a willingness to learn from everyone." Hmm thus, irrepective of whether a so called "master" is demonstrably expert *in the technology being worked on * or not, he is always a master .And the poor fool who does, in this case, know enough to "teach the master" is still an apprentice . The labels are rigid .Once someone becomes a "master", even a situation where the practitioner fails totally does not remove "masterhood" .Nice! You might think it would make sense to say that in this context where the new technology is relevant , the "master" and "apprentice" labels are reversed .Thus if an acknowledged superior programmer, say Linus Torvalds and I are pairing on creating a j2ee leasing system ,(I presume Linus will kill himself before he has to do this , this is all hypothetical, bear with me ), then, in the context of j2ee, logic would seem to indicate that *I* am the "master" (or "journeyman " or "Lord and Master" or "King Skeletor" .. use whatever labels you think cool ) and he is the "apprentice" (or "teeny weeny toddler" or whatever , as long as the label is "lesser" than mine ) .But no! Once a master always a master! it doesn't matter if you don't have a clue. The more significant question of course is "does this kind of labelling make any sense" ? hmmmmm.... Other paragraphs in the book identify mastery with "peer recognition" . There is a very subtle fallacy here .In other words,as per the book , Bruce Lee was not a master martial artist because he could fight better than anyone, but because he is *seen* to be a good martial artist by his peers . This confuses cause and effect . To see the fallacy embedded here, you can drop each condition in turn. Suppose Bruce Lee never became famous but retained as much of his fighting skill as he did . Is his "mastery" now non existent because he was not recognized ? Now let us try it the other way . If Bruce Lee was much much more famous amongst "peer" (of course it is not defined how these "peers" would be selected. maybe they are people with the same label from the guild? ) martial artists but could never demonstrate his skill in combat (an analogous situation would be a large number of book authors having no really significant software to back up their supposed "expertise") would he still truly be a "master" ? This labelling based on irrelevant attributes is the least part of the nonsense embedded in the book . More tragic is the notion of a whole set of practices to solve the "software crisis" built on this "labelling". Some of it is just common sense. Most of it is just misguided, intellectually unsound rubbish. Thus two separate concepts of "apprentice, journey man and master" --one, a rough measure of skill , and two, artificial "labels" used to recreate an outmoded economic system-- are conflated. Now let us look into the guild system from which this terrible classification scheme is borrowed . From the wikipedia entry for "Guilds" , " The guild was made up by experienced and confirmed experts in their field of handicraft. They were called master craftsmen. Before a new employee could rise to the level of mastery, he had to go through a schooling period during which he was first called an apprentice. After this period he could rise to the level of journeyman." This is almost a word for word description of the system McBreen wants us to replace present day practice with . Let us read further (the wikipedia article, not the book ) . " Apprentices would typically not learn more than the most basic techniques until they were trusted by their peers to keep the guild's or company's secrets. ". Duh ? secrets ? "Journeymen were generally paid by the day and were thus day laborers. After being employed by a master for several years, and after producing a qualifying piece of work, the apprentice attained the rank of journeyman and was given a letter which entitled him to travel to other towns and countries to learn the art from other masters." duh right . Does anyone else feel an increasing disconnect between the socio economic situation in the 14th century vs the present day ? Continuing to quote "After long periods of apprenticeship to a master" ( an apprentice had to pay for this privilege !What do you think of having to pay a "master" or "journeyman" to learn programming? ) "and after producing a qualifying piece of work, the apprentice attained the rank of journeyman and was given a letter which entitled him to travel to other towns and countries to learn the art from other masters." entitle him to travel ? a qualifying piece of work ? duh? Now let us give Mr McBreen's ideas the best possible interpretation and decide that your "level" in software needs to be demonstrated, say by showing samples of your code , maybe your contributions to open source, before you get a job. Will this miraculously improve the success rates of projects ? Not at all, because the problem is that the need for programmers is well above what can be met by the "truly skilled" minority . If less than say ,250 people worldwide are truly "masters" and say another 3,000 or so of "journeymen" status in, say, Tennis (going by ratings and being generous) why should there be a greater number in software development ? The difference is of course most people who play tennis do so because they at least minimally enjoy the game! To think that adopting some half baked scheme based on an ill defined outdated idea will raise these numbers(which is about the only way to "solve" the software engineering "crisis") is , at best , self deception and at worst .. hmmm ...let us not go there . In other words the "software problem" is economic, not technical in nature . Unless a compelling economic argument can be demonstrated for this switch to an out moded way of "improving" programmers , the only readers of this book are likely to be the mediocre programmers who fantasize about achieving "master" status and earn 250 k $/year (as Mc Breen reccomends) . Note to Mr McBreen . People who are that good are already earning 250k or more!! so what was the problem again ? Pete has written another half baked book on "Questioning Extreme Programming ".Far be it from me to say that XP needs no challenging , but this book does such a shoddy job, setting up and destroying a sequence of strawman arguments that wouldn't stand up in a high school debate,that you'd be surprised publishers would let this through their editorial board . Anyway if you want some good ideas on how to become a master programmer , go read this article . It has more wisdom in it than Mc Breen's Book of Mumbo Jumbo. So what is my take on this "apprentice,jouneyman,master" trichotomy ? Well I believe that in these days of universally available internet access and open source software any half way decent programmer is aware of where on the skill scale he (and any of his fellow programmers) falls , in any given context . Taking these "rating" labels seriously (for eg: on a site like Advogato,where you can see comments like " hey i became a journeyman while i wasn't looking !! wow ! " ) is, at best, like wearing your karate belt over your street clothes, to advertize your fighting skills . As Bruce Lee said ,"On the street, no one asks you your belt before they hit you " . Translated to the software development domain, that means "shut up and write code" .