Yes, industry typically pays more than academia. Yes, most molecular biologists cannot code and rely on bioinformatics support. Yes, biological data is often noisy. Yes, code in bionformatics is often research grade (poorly implemented, poorly documented, often not available). These are all good points that have been made many times more potently by others in the field like C. Titus Brown (http://ivory.idyll.org/blog/category/science.html). But they are not universal truths and exceptions to these trends abound. Show me an academic research software system in any field outside of biology that is functional and robust as the UCSC genome browser (serving >500,000 requests a day) or the NCBI's pubmed (serving ~200,000 requests a day). To conclude from common shortcomings of academic research programming that bioinformatics is "computational shit heap" is unjustified and far from an accurate assessment of the reality of the field.
From looking into this guy a bit (who I've never heard of before today in my 10+ years in the field), my take on what is going is here is that this is the rant of a disgruntled physicist/mathematician is a self-proclaimed perfectionist (https://documents.epfl.ch/users/r/ro/ross/www/values.html), who moved into biology but did not establish himself in the field. From what I can tell contrasting his CV (https://documents.epfl.ch/users/r/ro/ross/www/cv.pdf) to his linkedin profile (http://www.linkedin.com/pub/frederick-ross/13/81a/47), it does not appear that he completed his PhD after several years of work, which is always a sign of something something going awry and that someone has had a bad personal experience in academic research. I think this is most important light to interpret this blog post in, rather than an indictment of the field.
That said, I would also like to see bioinformatics die (or at least whither) and be replaced by computational biology (see differences in the two fields here: http://rbaltman.wordpress.com/2009/02/18/bioinformatics-comp...). Many of the problems that apparently Ross has experienced come from the fact that most biologists cannot code, and therefore two brains (the biologist's and the programmer's) are required to solve problems that require computing in biology. This leads to an abundance of technical and social problems, which as someone who can speak fluently to both communities pains me to see happen on a regular basis. Once the culture of biology shifts to see programming as an essential skill (like using a microscope or a pipette), biological problems can be solved by one brain and the problems that are created by miscommunication, differences in expectations, differences in background, etc. will be minimized and situations like this will become less common.
I for one am very bullish that bioinformatics/computational biology is still the biggest growth area in biology, which is the biggest domain of academic research, and highly recommend students to move into this area (http://caseybergman.wordpress.com/2012/07/31/top-n-reasons-t...). Clearly, academic research is not for everyone. If you are unlucky, can't hack it, or greener pastures come your way, so be it. Such is life. But programming in biology ain't going away anytime soon, and with one less body taking up a job in this domain, it looks like prospects have just gotten that little bit better for the rest of us.