There is no general rule even that all judicial decisions be published. In courts around the world, decisions are sometimes rendered orally and not transcribed. American courts make many "unpublished" decisions which have little or no precedential effect. And there is not as yet a general common law rule that reasons must be given for all administrative decisions.
Yet the problem here, as I understand it, is not that reasons have not been given. Indeed, written reasons are given to the parties. But they are circulated no further (absent freedom of information requests from interested lawyers).
Two reasons have been given, neither of which stands up to scrutiny. First, the decisions are said (correctly) to have no precedential value and the tribunals will publish those decisions that contain important propositions of general principle. But even if the decisions are not precedents, strictly speaking, the principle of consistency does apply to the tribunals. The Supreme Court of Canada has rejected the idea that inconsistent decisions are (without more) a basis for judicial review, but in the same breath it emphasized the importance of consistency:
While the analysis of the standard of review applicable in the case at bar has made clear the significance of the decision‑making autonomy of an administrative tribunal, the requirement of consistency is also an important objective. As our legal system abhors whatever is arbitrary, it must be based on a degree of consistency, equality and predictability in the application of the law. (Domtar).Indeed, its recognition of the importance of consistency has justified allowing decision-makers to hold internal discussions in the absence of the parties in order to discuss matters of principle (Consolidated Bathurst).
It is problematic if the tribunals reach different conclusions on similar or identical facts. Similarly situated individuals are entitled to be treated similarly. Publishing the tribunals' decisions would be a means of avoiding inconsistency. Good administration requires no less. Otherwise, similarly situated individuals would potentially be subject to different decisions depending on which panel of decision-makers they happen to appear before.
The tribunals' undertaking to publish decisions that they consider to be significant does not address this concern. If all decisions were published in a searchable electronic database, as is the general norm with judicial decisions, parties could decide for themselves what cases are relevant or whether patterns can be discerned in the decisions of the tribunals. The tribunals would effectively outsource to counsel the process of distilling principles from its decisions, a much more efficient arrangement.
The second reason given is that the cost of translating the decisions into both official languages (French and English) is too high. I doubt that this argument would survive judicial review. The tribunals are not required by legislation to publish only a selection of their decisions. They should therefore exercise their powers in conformity with the principles of procedural fairness.
Fairness is imperilled by selective publication in at least two ways. First, government agencies are repeat players before the tribunals and will have access internally to large banks of decisions. This creates an inégalité des armes between the government and the individual (especially if the individual is self-represented). Second, the tribunal members will presumably try to achieve some sort of consistency in their own decision-making and in doing so take 'official notice' of their past decisions. Doing so without revealing these decisions to the parties concerned is unfair.
The tribunals really should publish or be damned.