Expressing financial information as an SEC XBRL financial filing neither adds to nor removes any subjectivity or professional judgment required to express financial information within a financial report.
Now, there are inconsistencies and ambiguities in the US GAAP taxonomy and gaps in SEC Edgar Filer Manual (EFM) rules which can make exactly how to express such information more challenging and which can lead to different correct conclusions which impact compatibility of such financial reports.
There seems to be three moving parts at play:
Fundamentally, there is a difference between how SEC public companies report under "US GAAP" and "the US GAAP taxonomy".
How XBRL works is not a mystery. How XBRL works is articulated in the XBRL 2.1, XBRL Dimensions, and XBRL Formula technical specifications. There is zero room for subjectivity here. How could subjectivity exist and we have interoperable software? The entire point of a global standard such as XBRL is to agree as to how things work and document this information within a technical specification so that interoperable software can be created.
My personal assessment which is based on the evidence that I see is that there is zero subjectivity and professional judgment relating to expressing information such as an SEC financial filing using XBRL other than the subjectivity and professional judgment which has always been there with paper-based financial reports.
An SEC XBRL financial report is a definitive, discrete, finite set of semantic objects, properties of those semantic objects, relations between semantic objects, and properties of the relations between semantic objects. These semantic objects, relations, and properties fall into categories which is helpful. These semantic objects, relations, and properties are expressed using the XBRL technical syntax which has no room for subjectivity or professional judgment.
But, the objects themselves are different than the informtion expressed by those objects.
US GAAP has semantics, such as assets = liabilities and equity; the income statement foots; authorized shares must be greater than or equal to issues shares; issued shares must be greater than or equal to outstanding shares; etc. There is a finite number of these disclosure rules which exist.
Is US GAAP totally unambiguous? No. However, nor is US GAAP arbitrary or random. Users of US GAAP have choices, they have options. Subjectivity and professional judgement need to be exercised when applying US GAAP to financial disclosures. But not everything is subjective or requires professional judgement. For example, it would be rather hard to dispute "assets = liabilities and equity" or the balance sheet balances.
In fact, US GAAP limits judgement and discretion. The conceptual framework does this by (per the FASB Special Report, The Framework of Financial Accounting Concepts and Standards (1998):
Does the US GAAP Taxonomy correctly mirror the semantics of reporting financial information under US GAAP? It should, totally. But it does not. The US GAAP Taxonomy has inconsistencies, the US GAAP taxonomy and SEC filing rules are not totally unambiguous. Should the US GAAP taxonomy and SEC filing rules be unambiguous? Sure. Will they ever be? Doubtful.
One of the ways ambiguity of the US GAAP taxonomy reveals itself is options as to how you model something like the classes of property, plant and equipment; do you model those as concepts or do you model them as the [Member]s of an [Axis]? While this can be done in different ways sometimes; other times you do not get to pick between the two options due to modeling choices made in other places within your SEC XBRL financial filing.
Because the US GAAP taxonomy and things like the XBRL US consistency suite are incomplete as to 100% of what is required under US GAAP disclosure rules, two things must happen.
First, reporting entities need to express additional semantics and then verify that their SEC XBRL financial filing is correct using automated processes assisted by software applications. Second, for all unautomatible verification tests or for all automatable verification tests for rules (which would allow automation but for which no rule was created which would have allowed automation of the rule) do not exist; manual verification steps/tasks must be performed to verify that the SEC XBRL financial filing is a true and fair representation of the financial information of the reporting entity (correct, complete, consistent, accurate, has fidelity, has integrity).
But, all things considered the semantics of the objects, relations, properties of an SEC XBRL financial filing need to work according to US GAAP per the technical rules articulated by the XBRL 2.1, XBRL Dimensions, and XBRL Formula specifications.
So, many times discretion as to what must be reported or how it is reported can be subjective and requires professional judgment; the actual expression, that specific financial report, must be modeled correctly per the technical rules.
Comparability between different specific reports or of such models is a different issue. While each individual report model must be correct, different reporting entities have different modeling options. Want better comparability? Tighten up the US GAAP taxonomy, the EFM; make things more consistent, more unambiguous, and be as explicit as possible so that computers don't need to imply information and potentially imply something incorrectly. But you will still never achieve full comparability.
Comparability can only be as good as the comparability required by US GAAP. That issue cannot be fixed by changing the US GAAP taxonomy or SEC filing rules. The US GAAP taxonomy and EFM rules can be 100% consistent, 100% unambiguous, and 100% explicit; and there will still be comparability issues.
Getting better comparability than offered by what existing US GAAP requires today will require for US GAAP to be changed. And that is yet a different issue. Other technical issues exist also such as the need for common metadata such as SIC codes and industry sectors.