Re: Is open source a healthy choice? (July 25)

This is in response to some of the recent letters that appeared in response to the above article.

In open source software (OSS) projects, the many eyeballs argument describes a form of peer review. There is no question that peer review is one of the most effective mechanisms for finding defects in software. But the evidence that exists suggests that not all OSS projects do peer review (one survey shows only 9 per cent of OSS developers agreeing that all of their code is peer reviewed). So, even if it is possible to do peer review, that does not mean that it actually happens. And those OSS projects that do would not necessarily have an advantage over closed source software projects: many closed source projects do have very effective peer reviews.

In addition, many OSS projects do not have good pre-release testing practices, and in software engineering we know that peer reviews and testing do not find the same defects – so there is a whole class of detectable defects that get shipped to end-users in OSS. The users will eventually find these defects too and some users will report them. But in health care we simply cannot take these kinds of risks. OSS projects need to have effective testing practices in place.

It is true that it is easier to measure quality with OSS. But because it is possible does not mean it happens. There have only been two serious studies thus far that attempt to measure the quality of OSS (for Apache and FreeBSD). Apache was found to have lower post-release quality than products from a telecommunications company and FreeBSD had mixed results when compared to the same commercial products.

We have done a systematic review of all of the published evidence – and that is what we found:

The fact is that many people and SMEs adopt OSS because they believe that it is free (as in “free beer”). But at the enterprise level it is essential to pay for support for OSS – so really it is not free. Is the total cost of ownership higher or lower for OSS? I have seen good cases made for and against. Such cost analysis models tend to be sensitive to the assumptions and parameter values – so this is something that would have to be evaluated on a case-by-case basis.

There are many closed-source vendors that provide great support for their products. And if you don’t like the support, or if you do not like the product, then go to a competitor. That’s how the market works. The competitor may be OSS or closed source.

Not all OSS products have commercial support, however – which is a practical problem, even if the application is attractive. OSS does fill a need, but one has to be careful to separate the claims from the data. The wise will make adoption decisions based on the data.

Khaled El Emam (PhD)
Associate Professor, Faculty of Medicine, University of Ottawa
Canada Research Chair in Electronic Health Information


Re: So smart, it’s a crime
(Aug. 19)

Some criminals use technology to break laws. To “level the playing field” let’s legalize law enforcement to use technology to break existing laws. Hmm, some criminals kill people to commit crimes. To level the playing field let’s legalize law enforcement to kill people to prevent crimes.

P. Paulson
Newmarket

Re: So smart, it’s a crime
(Aug. 19)

“Here Here,” as the MPs like to say!Norma Hewitt-Lendrum


Re: BIOS on the way out (Aug. 25)

Certainly it is encouraging to hear that BIOS’s days are numbered. Extensible Firmware? If I understand the concept then I wonder if it is a major technological breakthrough. Wasn’t that how Apple delivered its Mac OS back in the good old days? I don’t know if they still do. Great idea, though, especially if Intel can come up with a model which permits cramming all of a bloated OS like XP into firmware.

Keith A. Ujvary
SIDUGEC Software Technology
Gibsons, BC


Letters to the editor must include the writer’s name and company name along with an e-mail address or other contact information. All letters become the property of ITBusiness.ca. Editors reserve the right to edit submissions for length and content.

Share on LinkedIn Share with Google+