BLOG: Digital Financial Reporting
This is a blog for information relating to digital financial reporting. It is for innovators and early adopters who are ushering in a new era of digital financial reporting.
Much of the information contained in this blog is summarized, condensed, better organized and articulated in my book XBRL for Dummies and in the three documents on this digital financial reporting page.
The quality of public company XBRL-based financial reports submitted to the SEC continues to improve. The last measurement of a specific set of fundamental accounting concept relations showed that 58% of all such public company financial reports were consistent with expected basic accounting relations. That consistency rate has increased 5% to 63%. The average number of inconsistencies per financial report (for my set of relations) has decreased from .7 per report to .6. The total number of inconsistencies dropped by 626.
This is the summary of information by software vendor and/or filing agent (generator):
There were 11 software vendors/filing agents that had a 5% increase in quality or greater as compared to my prior analysis in April 2015.
This graphic shows comparison information for the 2013 10-K season, the 2014 10-K season, and this most current analysis:
Competition is a good thing. Public companies have plenty of options when it comes to software vendors and/or filing agents for helping them create their digital financial reports. These choices will help push quality even higher, I believe, and will help the financial reporting supply chain achieve what other domains such as health care and electronic medical records are trying to achieve.
Want to understand how to improve your consistency with these basic fundamental accounting concept relations? This chapter of Digital Financial Reporting provides information about the fundamental accounting concepts. This page on my blog provided very detailed information on these accounting relations. If you want to understand the bigger picture, I suggest these two resources: Digital Financial Reporting Principles and Knowledge Engineering Basics for Professional Accountants.
I will measure quality again in three months. I predict that there will be another healthy increase in quality, that over two thirds of all public company XBRL-based financial filings to the SEC will be consistent with all of these basic accounting relations, and that at least four software vendors/filing agents will have 80% of their financial filings or more completely consistent with these basic accounting relations.
The article, What is cognitive computing? IBM Watson as an example, says that cognitive computing is the third era of computing:
If we consider that the first era of computing that the one of tabulating systems (1900), the 2nd the one of programmable systems (1950), cognitive computing is the 3rd era of computing.
It is a mistake to overstate the capabilities of something. It is likewise a mistake to understate the capabilities of something. Philosophers, theologians, and academics like to debate things simply for the sake of the debate. Business professionals, me included, tend to take a different approach. We like to use practical tools that actually work. Me personally, I don't see a computer having a conscience any time soon.
People love buzz words. Early in my career one buzz word was "client server". Few people seemed to know what client server was, but everyone felt they needed one. Artificial intelligence has been a buzz word for quite some time. Some of today's buzz words are: Big data. Smart data. Smart machines. There are even people who track buzz words. I tend to take this persons view on buzz words:
Not all buzzwords are bad. Some actually convey an idea, a concept, or a valid technology. However, some exist to confuse, distort, and empower those who don't want you to know what they're talking about.
Many people tend to miss a really important point when it comes to XBRL. They look at XBRL as "tagging" and they equate XBRL to the "barcode". But, turn the equation around. What would it mean if all financial information were properly bar coded? What could you do? What would that enable?
Think about this: What would happen if you took financial reports and rather than reporting information in an unstructured format that only machines could understand (because the information is unstructured or more accurately structured for presentation); but rather reported information was in a structured format both humans and machines could read and understand? What exactly would that mean?
Well guess what. We will get a chance to see exactly what that means. Financial information is being reported to the SEC by public companies in the structured format XBRL. The quality of that information is being dialed in. Professional accountants are understanding more and more about how digital works. Professional accountants are building prototype financial report ontologies and suggesting how ontologies can be used to improve financial reporting. Professional accountants are coming up with ways to work with technical artifacts so that they can employ these useful technologies on their terms.
Cognitive computing is the simulation of human thought processes in a computerized model.
Notice the word simulation. Computers cannot think, they are dumb beasts. But these dumb beasts can be made to mimic the human thought process, if you understand how to harness the power of a computer. (If you don't understand what it takes to harness the power of a computer, read Zeroing in on the Holy Grail of Global Standard Financial Reporting.)
Computers do not create the magic. Skilled craftsmen who wield their tools effectively are what create the magic. Computers simply follow instructions.
This article, Cognitive Computing And Semantic Technology: When Worlds Connect, points out something that is really important in two key statements in that article:
For cognitive computing to achieve its promise we need a thick metadata layer that incorporate semantic tagging formats.
A lot of the focus is on machine learning, especially as things move to really analyzing and building explicit knowledge models, but other areas that should be included in the cognitive computing mix include constructive ontologies and constructive knowledge modeling, whether it's done by groups or individuals or crowd-sourced in the case of the semantic web.
So, what is not in dispute is the need for a "thick metadata layer" in order for the computer to be able to perform useful work. But what is sometimes disputed, it seems, is HOW to get that thick metadata layer. There are two basic approaches:
- Have the computer figure out what the metadata is: This approach uses artificial intelligence, machine learning, and other high-tech approaches to detecting patterns and figuring out the metadata.
- Tell the computer what the metadata is: This approach leverages business domain experts and knowledge engineers to piece together the metadata so that the metadata becomes available.
Now, I understand many things about how computers work. Not remotely everything, but a lot. If there is an error in my understanding of what computers can achieve, it would tend to under estimate what they can achieve rather than over estimate.
As a professional accountant, I understand that the probability of a computer starting from scratch and using the most sophisticated technologies and approaches available today and creating any useful metadata is very close to zero. However, the more manually created metadata that a computer has to work with, the higher the probability that the computer would be helpful in correctly figuring out financial reporting metadata.
So what I am saying is that humans are going to have to prime the pump and get quite a lot of metadata pieced together. Then at some point and for some things, computers can effectively be used to contribute more metadata. And so, this is not an either-or question. Both approaches can be used effectively and contribute to what is needed to realize the potential offered by cognitive computing. I am also saying that there are no short cuts.
Can cognitive computing work and have an impact on financial reporting? No doubt. Work practices of professional accountants will be changing over the coming years in very big ways. 1 year? 5 years? 10 years? 15 years? Hard to say. So keep Gartner's Hype Cycle in the back of your mind.
Computers assisting professional accountants in correctly representing financial reports digitally will cause high-quality financial information to be available for analysis by investors and regulators. Everyone in the financial reporting supply chain will benefit from the meaningful exchange of financial information in machine-readable formats.
XBRL was never about "tagging" or "barcodes". XBRL is about all the possibilities that are enabled if information can be successfully exchanged between business systems. While financial reporting is leading the way, in particular XBRL-based financial reporting by public companies to the U.S. Securities and Exchange Commission, other reporting schemes will likely benefit from the SEC's bold experiment. Business professionals will build their own Watson-type systems for way less than what IBM paid to build Watson.
Concept computing will contribute to changing how financial reports are created similar to how CAD/CAM contributed to how blueprints are created and how the design supply chain interacts.
Just like many other things a taxonomy or ontology has a life cycle. The paper Towards ontology evaluation across the life cycle explains the problem of not understanding that life cycle and not being able to evaluate the quality of an ontology:
Problem: Currently, there is no agreed on methodology for development of ontologies, and there is no consensus on how ontologies should be evaluated. Consequently, evaluation techniques and tools are not widely utilized in the development of ontologies. This can lead to ontologies of poor quality and is an obstacle to the successful deployment of ontologies as a technology.
The paper points out that there are five aspects to the quality of ontologies which need to be evaluated:
The paper provides this diagram of the different stages of the taxonomy/ontology life cycle:
This is a list of the stages which are explained in the document:
- System design
- Ontology design
- Ontological analysis
- Requirements definition
- System development and integration
- Ontology development and reuse
I missed this in my chapter on Knowledge Engineering Basics for Professional Accountants, but will add it in the very near future.
I have been fiddling around with SECXBRL.info for quite some time. 28msec is pretty liberal about making the service available. The DOW 30 information is available for free. If you sign up, which is also free, you basically have access to all information for all the companies who file 10-K and 10-Q filings except for trusts and funds.
I create some documentation a while back which shows how to create: (some of these might need updating)
To make this even easier to understand, I create the following example queries and reports which also includes 28msec's new spreadsheet-to-report type queries. Here are those.
- RSS Feed of queries comparing IBM and Microsoft: A set of 6 queries which compares IBM and Microsoft.
- Compare IBM and Microsoft: Demonstrates how you can embed queries within a web page and render multiple queries together on one page. (This uses the queries in the RSS above and simply organizes the queries in an HTML page using iframes.)
- RSS Feed of 27 fact queries: Set of 27 different fact queries which returns results in a common fact table format.
- Compare across entities (RSS Feed): Set of queries which demonstrate how to compare information across entities. This uses the spreadsheet-to-report to return human readable results. (also available in RDF)
- Compare across periods (RSS Feed): Set of queries which demonstrate how to compare information for the same entity across periods. Also uses the spreadsheet-to-report human readable format. (also available in RDF)
This stuff is really getting useful!
A little application called Scratch which was created by MIT shows those who have a little imagination where digital financial reporting is headed;
If you need a little more help getting your head around that idea, the document Understanding Blocks, Slots, Templates and Exemplars helps business professionals begin to understand how digital financial reporting will eventually work.
Software engineers use patterns to implement software. A financial report is not one big thing, it is a many little things. The notion of block, slot, template, and exemplar help give software engineers things to work with to implement sophisticated software that is easy for business professionals to use.
Imagine accounting professionals interacting with this:
Technical? Absolutely that is technical. Just not information technology-technical; it is accounting-technical. And that is exactly what professional accountants shoud be working with, accounting related stuff.
Can anyone explain to me why digital financial reporting software cannot work has I have described in the document? Everything is glued together with business rules understood by professional accountants. Software does more to help accounting professionals get things right, rather than give them more opportunities to make mistakes.
Why can't this work?