IPWatchdog posted this article recently. It's well worth a read for clients working in the software space and for practitioners that prepare specifications covering software.
I will jump to the "lessons". It appears that the description should set out how a general purpose computer is transformed into a "specific use computer". This means, to quote the author, "...applications should include an abundance of disclosure, diagrams and images providing as much tangible description of everything you mention in the claims, particularly if you are going to rely on means-plus-function claiming as is frequently done in the computer software area." This includes algorithms. It is the algorithms that will perform the necessary transformation of the computer. Without them, a judge or examiner may fail to see how the computer is transformed, even thought that would be clear to a person "skilled in the art", the intended addressee of the patent.
All of this extra work seems to be required to convince US judges and, yes, AU examiners that exploitation of the invention results in the required transformation of the computer.
In my recent experience, the attitude of AU examiners has been simply to object based on lack of subject matter without careful consideration of our landmark case: National Research Development Corporation v Commissioner of Patents (" NRDC case")  HCA 67. In a sense, this attitude has not been helped by those applicants who have attempted to "push the envelope" in cases such as: Grant v Commissioner of Patents  FCAFC 120 (18 July 2006), Peter Szabo and Associates Pty Ltd  APO 24 (5 May 2005) and the almost ridiculous attempt in Invention Pathways Pty Ltd  APO 10 (21 July 2010). I have commented on these and others here. Examiners invariably refer to the latest case, disregarding hierarchical precedent. Where the latest case is something like the Invention Pathways case, the applicant can be prejudiced.
Unfortunately, this all ends up costing more for clients. In fact, it would not be unusual for a patent application for a software product to cost up to twice as much as a patent application for even a complex mechanical product.