Creating something like an accounting oracle machine certainly seems conceivable. Computability theory and computational complexity theory seem to provide information about some of the possibilities here. The information in an area of knowledge such as the area of accounting, reporting, auditing, and analysis can be organized using the Cynefin Framework. The information is not "random" or "unorganizable".
The hard part about creating something like an accounting oracle machine is not the computation part; the hard part is more about pulling all the machine-readable rules together effectively and efficiently. This is expensive and time consuming and is not something that any one organization can really do.
But a coordinated group could pull this off and do so effectively and efficiently. The tools of the information age can be brought to bare to help make this happen.
Building an accounting oracle machine is about giving accountants the capabilities to organize the information represented by logic gates that make up the machine-readable knowledge graph that drives the oracle.
Do issues exist when you try and create something like an accounting oracle machine? Certainly. Not every piece of information will be disputed, for example, the accounting equation (“Assets = Liabilities + Equity”) is really not disputed. But there can be areas where multiple perspectives jostle for prominence, the leaders of different factions of each perspective will argue with one another, and dissonance will rule the day.
However, a good conductor can get the orchestra into harmony.
Achieving this harmony can be hard in some cases, but in the area of knowledge such as accounting, because it is an organized profession with clear rules in most cases; something like an accounting oracle machine is quite possible.
An accounting oracle machine will be made up of two things: rules, engine (to process rules). The rules will be created by knowledgeable business professionals and will be declarative as per the Business Rules Manifesto. The rules format will be some global standard, XBRL can work. The engine will be one of the three primary problem solving logic implementation approaches. Here are the beginnings of such an accounting oracle machine:
To the untrained eye, all this seems impossible. When using proper sensemaking techniques, all this is really pretty straight forward to understand. Accounting is a very straightforward mathematical model. For the things that can, and should, be put into something like an accounting oracle machine here is the breakdown; 80% anyone can create, about 18% will be easy for professional accountants to create, about 1.8% will be hard for accountants to create, and .2% will be a complete struggle. About 80% of the effort for accountants is understand WHAT should be defined in such an oracle and 20% of the effort will be doing the work.
