Letters

I've been deeply involved with MPEG for several months now, so your article on MPEG-1 playback programs caught my eye (LJ May 2001). I found several factual errors about MPEG in the article.

First, MPEG-2 video does not necessarily have better video quality than MPEG-1: the differences between the video portions of the MPEG-1 and MPEG-2 standards are fairly minor. An MPEG-1 file with a 720 x 480 frame size, compressed to 6 Mbit/s is legal, and will be very similar in appearance to a 720 x 480 6 Mbit/s MPEG-2 movie.

Second, the author states that “not all MPEG-1 files are entirely `compliant”'. The MPEG standards define what a compliant decoder is expected to be able to handle, but not how a compliant encoder works. In other words, a compliant decoder is not one that is “tolerant”, but is instead one that adheres closely to the letter of the ISO 13818 standard, so as not to be surprised by the output of a novel (but legal!) encoder. Unfortunately for free software authors, these documents are expensive: the basic set of MPEG-2 PDFs (ISO 13818-1 through -3) costs $424, and the complete set is $1,390 (from http://webstore.ansi.org/). It's little wonder that commercial MPEG products far outstrip free ones' capabilities, in nearly all cases.

Third, the author states that an “MPEG-1 audio stream...is an MP3 file”. This is not always true, and is not even likely. MPEG-1 defines three different audio encodings (or “layers”). Layer I audio is the most basic, but it isn't used very often. Layer II is the most common: video CDs and MPEG files you download from the Internet almost always use Layer II audio. Layer III (aka MP3) is optional, so most MPEG decoders don't include support for it. Since few decoders support Layer III audio, most encoder creators also don't bother including support for it.

—Warren Young

Just a note to let you know that the certified sword cuts both directions (Editorial focus, LJ May 2001). As CIO of a multimillion dollar corporation it is my job to not only run the IS department but to hire and fire employees. One of the first questions I ask is if the prospective employee has any certifications, if so I politely tell them “I'll let you know if we can use you” and promptly throw their application into the garbage can.

Before you ask, no I am not certified. I have however taught pre-MCSE classes at Unisoft Institute of Technology in Houston and was horrified to learn that I had to teach the way the test worked, and the way the real world worked (pre-MCSE in this case really meant A+ and Networking certification). This effectively meant that I held two classes in one, which to say the least was difficult. Additionally, long ago before computers were my profession I was an ASE-certified mechanic. Since I have passed 10 ASE certifications I can tell you that they are just as much a joke as computer certifications. I quickly realized that even holding all the certifications I did, and after graduating from a top automotive technical school with a 4.0 GPA and Alpha Beta Kappa National Honor Society, that I was not a very good mechanic.

To me, being certified means that the person does not have enough knowledge or experience to get the job on their own merits and hopes that this piece of paper will help them, and it does not. In my experience the only time certifications help you is when you are applying to a business where the person responsible for creating hiring policies is not a real computer technician.

I am forced to deal with MIS and IS degrees from recent college graduates as well as a plethora of certifications on a regular basis. Unfortunately, I have found that the people who have neither a degree nor a certification but who have been working with computers for ten years are much better equipped to handle the job. At least if they are inexperienced I can teach them the way things really work instead of attempting to retrain them after they have their degree or certification.

—Allan Hall

I very much enjoyed the April 2001 Linux Journal article “Linux on Carrier Grade Web Servers”. You did a nice job of describing the software choice, hardware environment and test results. I look forward to future articles discussing the other LVS implementations (direct routing and IP tunneling) and comparing their stability and performance with that of the NAT implementation.

Thank you.

—Bill Landahl

I enjoyed your articles on Linux Certification (LJ May 2001). I thought the “real-life” experience was very telling, although perhaps toned-down a bit to protect the vendors.

Here's my thoughts on what I read:

We can earn an extra $10K per year by becoming certified? Really? Who can? New grads? Having worked on, supported and/or maintained SVR4, AIX, HP-UX and Solaris for twenty years, I can't imagine getting another $10K just because I had some Linux certificate.

I looked at some of the questions from Red Hat's and Sair's study guides and tests. What a crock! “What's the fdisk type code for a Linux swap partition?” Who cares?! Look it up by typing l to list the types. Forgot that command? Type ? to list them all. Better yet were the impossible to understand questions and answers on Sair's test. Their “correct” answer for wc -l * is that it returns the total number of lines in the files. Gee, my experience is that it shows the line count for each file, followed by the total, but that answer isn't available. Yes, they want the entire Linux community to help improve the tests, but if they can't get the simple things right, I'd hate to see how they do with the hard topics.

Finally, the exam companies could learn a good lesson from the FCC and ARRL. The amateur radio exams are also multiple choice, but they are composed of a certain number of sub-elements. Each sub-element has a number or required topics. Each topic has a number of published questions and answers, with references to the rules and regulations. The effect is, the actual exam might have only 25 questions, but those questions are pulled from a pool of several hundred, and each critical element is covered.

—Mike Hall