News from PANUG/BizNix - July 3, 2002 http://panug.org - http://biznix.org LINUX NETWORKING The Linux Networking class is nearly full. We're limiting the class size to 12 people. We have 2 seats left. The 3-day class is held on non-consecutive days so students can digest and practice what they learn instead of having to digest three days of intensive details in one shot. The class dates are July 11, 18, and 25. For information about the course contents go here: http://alcpress.com/training/courses/linuxnetworking.htm The course costs $325. To signup, call 503-635-6370. BUGGY SOFTWARE COSTS $59.5 BILLION by Ed Sawicki - Accelerated Learning Center / Tailored Computers The National Institute of Standards and Technology has issued the findings of a study it conducted on the cost of buggy software. The cost, $59.5 billion, is borne mostly by consumers of the buggy software. They say "few products of any type other than software are shipped with such high levels of errors." The study found that over half of all bugs were discovered well after the code was written - mostly when the software was already in the hands of users. Clearly, this suggests a lack of testing but the NIST report says that about 80% of software costs are spent on testing. Something is really wrong. Why isn't testing finding the bugs? NIST blames this mostly on growing complexity. Popular software is now measured in millions of lines of code. Both programmers and their testing tools are challenged when code size grows very large. But there are other reasons as well: 1.Marketing strategies - this is pretty broad. It includes such things as time-to-market pressures that force programmers to take shortcuts to meet agressive schedules. This helps explain why, for example, a company can hire brilliant programmers but still ship awful products. 2.Limited liability by software vendors - since software vendors are somewhat insulated from lawsuit, there's little incentive to reduce bugs. 3.Decreasing average product life expectancy - vendors keep profits high by eating their young and forcing consumers to upgrade to the next version of buggy software. Unfortunately, each new version may fix some old bugs but introduce new ones. The study does not mention vendors, hardware/software platforms, or software categories as the worst offenders. This is unfortunate because readers could easily get the impression that the entire industry suffers from these problems equally, which is not the case. The problems mentioned above are self-inflicted. Vendors choose to have these problems. A vendor chooses to keep growing a program until its complexity is such that it can no longer be properly maintained. No programming class or book in the world instructs programmers that "bigger is better". In fact, the reverse is true. Every Unix programmer has been taught that "small is beautiful". The problem of buggy software was created by the way the commercial software world does business. The solution must come from vendors making business decisions. This is not likely to happen. Buggy software means high profits. The NIST report is here: http://www.nist.gov/public_affairs/releases/n02-10.htm