It was early 1998, and the commercial side of Free and Open Source Software (FOSS) was beginning to flourish. Enthusiasts and researchers had embraced FOSS some time before, but while some companies were using FOSS, and some companies were making money from FOSS, most companies were still going through a “trial” period. Most commercial companies did not think that FOSS was “for real”, and they certainly did not treat it with the seriousness that people do today. It was at this time that a few of the member companies of Linux International started thinking about certification efforts for FOSS professionals.
The specific problem that we foresaw had to do with small companies that needed FOSS professionals, but who were not familiar enough with FOSS or even commercial Unix systems to be able to hire good staff. As demand for FOSS staff grew, we needed a good way of certifying that a potential staff member applying for a job knew more than just how to spell “FOSS”. This was particularly important in the area of systems administration.Imagine the manager of a small bank or office building needing to hire a FOSS professional for their Linux systems, but not knowing what questions to ask during the interview. Of course they would look at the candidates resume and talk with their previous employers, but there should be something more they could add to the mix to determine the quality of the person in front of them.
While certifications are not perfect, a good one can help to separate the wheat from the chaff. Certifications already existed, of course. Companies like Microsoft, Cisco, Oracle, Novell and other major companies had certification courses for their own products. Likewise some of the larger distributions such as Red Hat and SuSE were developing certification courses for their distributions. However these certifications were typically specific to a specific product or set of products. There were no certifications that were “generic” to FOSS systems administration, and many people in FOSS systems administration did not want to get certifications for each of the many distributions of GNU/Linux that exist, particularly since they correctly perceived that much of the knowledge would be shared across distributions through the use of common subsystems. “bash syntax is bash syntax”.
A small group of people formed to discuss the attributes of a certification organization. We felt that certification should be separated from training, so that people could attain the knowledge for the certification any way they desired. Training could be in the form of lecture-lab, or reading a book and practicing, by reading information on the Internet, or learning from “the school of hard knocks”. Separating certification from training also helped reduce the perceived desire to change the certification in order to drive more sales of training. Others in the fledgling LPI organization felt that LPI should be non-profit organization. In the spirit of FOSS, while we knew that certification would cost money to implement and administer and there would have to be paid staff, we also felt that certification should not be part of a profit-making organization. We wanted the certification to be a world-wide and respected. Unfortunately at the time of LPI's founding, there were some certifications that were so poorly done and administrated that they had become an industry joke and therefore useless to everyone. LPI wanted to avoid these issues.
Out of these needs LPI formed certain strategies. The exams would be developed with input from the FOSS community. The community would help determine the scope of the tests overall, and the scope of individual tests. For example, what should be the skill-set of a junior systems administrator as opposed to a more senior person? This is known as the “Job Task Analysis” (JTA), and for the first JTA, performed in April of 1999 over 1400 systems administrators from around the world participated. These participants listed the types of tasks they needed to do as a junior or senior systems administrator. The JTA surveys were evaluated, analyzed and formulated into objectives for each of the projected tests.
Today the JTAs continue to be performed for the different levels of LPIC testing at various times, with the latest round being performed in 2008 and 2009. Once you have determined the skill-set, what sets of questions should you ask the candidate to determine if they really have that knowledge? Once you have determined the types of questions to ask, exactly what questions should be asked, and what would a proper response be? Again, LPI looked to the broad community of FOSS systems administrators to help formulate a pool of questions and answers for their tests. LPI used techniques such as Applied Psychometrics, a science taught at Fordham University in New York, in test construction. Psychometrics looks at the results of tests to determine if the tests results are valid, and not thrown off by having one question influence the answers to others in either a positive or negative fashion. Psychometrics also evaluates if test questions are “too hard” or “too easy” by using statistical analysis applied to human actions. Finally, it was determined that the exams should not be distribution specific, and would be delivered in a way that would make the exams available world-wide at the lowest possible cost.
Moving forward A board of directors and a set of officers were formed, and while some early companies helped with various funds to help get the organization off the ground, a lot of the early work was by volunteers who saw the need and wanted to help. Much work was done by mailing lists, with the occasional face-to-face meetings at events such as Linuxworld in San Francisco (with all-day meetings hosted, complete with lunch, at IBM's offices). While it sometimes felt that we were moving forward at the speed of molasses on a cold day, we did make steady progress and the day came when we had the first of the tests ready to administer.
Chickens and Eggs: My involvement with LPI As I said at the beginning of this blog entry, the members of Linux International were fully behind the creation of LPI in 1998. It was even suggested that LPI become a branch of Linux International, but I felt that this would be a bad idea. It would be better, I felt, to have a separate organization focused solely on certification, but I did volunteer my time, effort and experience to help move LPI along. Then one day in late 1999 the fledgling LPI had a slight problem, one that I am proud to say I helped to rectify. In order to verify that the exams were psychometrically accurate, fair and worthy of the LPI brand, LPI had to have a certain number of the exams performed and graded. But without the brand, there was nothing to induce a candidate to pay the money to take the exam. No inducement, no exams taken. No exams taken, no way to prove the validity of the exams. This was the classic “chicken and egg” problem. LPI thought about offering some of the exams “for free”, but even doing that would cost money that LPI did not have. I sat in the meeting with the LPI staff and board as we wondered how this knot would be untied. Finally I asked the question “How much money are we talking about?” The answer was approximately one thousand dollars.
So on August 9th, 1999 (two days after my 49th birthday, and exactly thirty years after my start in computer science) I pulled out my personal checkbook and wrote a personal check for one thousand dollars to LPI for funding the first of the tests. I could not sit back and watch such an important issue die for lack of such a small amount of money. After that point LPI was able to move forward with tests that had legitimately met the standards that LPI had set for them. Others (both individuals and companies) pulled out their checkbooks too that August day in 1999, but for my action I later received a plaque from LPI which I value very highly to this day:
“The Linux Professional Institute is proud and appreciative of the participation of Jon 'maddog' Hall as a Platinum Individual Charter Sponsor of LPI, and its efforts towards advancing the use of Linux in Professional computing. - August 9th, 1999.”
And “advancing the use of Linux in Professional computing” is why LPI keeps moving forward, and why I am proud to be associated with it.