One of the many statistical models listed was MDL, minimum description length. Proving that no shorter description exists is not computable. The long list of other models are computable approximations of MDL. Some work better than others, depending on the data.
I am sure you are aware that statistical modeling is the same as compression. I don't disagree with you on this. All compressed codes can be thought of as programs. For example, LZ77 is a programming language with 2 opcodes, one to output a literal string and one to output a substring of the previous output. -- Matt Mahoney, [email protected] On Wed, Jun 10, 2026, 4:50 PM James Bowery <[email protected]> wrote: > > > On Wed, Jun 10, 2026 at 2:06 PM Matt Mahoney <[email protected]> > wrote: > >> ... >> As for model selection criteria, there are a lot of them for the same >> reason there are lots of different compression algorithms. The general >> problem is not computable. >> > > No one ever demands that a theory be the best of all POSSIBLE theories > relative to a given set of data. > > The fact that "uncomputability" is brought up with regard to executable > codes as opposed to statistical codes (Rissannen) for model selection is > base sophistry to avoid the clarity of algorithmic codes. Computability is > not and never has been the point regarding model selection. > *Artificial General Intelligence List <https://agi.topicbox.com/latest>* > / AGI / see discussions <https://agi.topicbox.com/groups/agi> + > participants <https://agi.topicbox.com/groups/agi/members> + > delivery options <https://agi.topicbox.com/groups/agi/subscription> > Permalink > <https://agi.topicbox.com/groups/agi/T5b182c3fd14f61e9-M2ce6aaa05c587de3b7c6d878> > ------------------------------------------ Artificial General Intelligence List: AGI Permalink: https://agi.topicbox.com/groups/agi/T5b182c3fd14f61e9-M57fb96d4069030b99d37e755 Delivery options: https://agi.topicbox.com/groups/agi/subscription
