Jon Stewart Don't Know Squat About Data

This past Wednesday (Mar 27, 2013), the Daily Show included a segment on the backlogs within the VA and found the cause to be

• The reliance on paper to transfer medical records between the DoD and the VA

So far, so good.  Good investigative reporting.  If they had only left it at that and allowed viewers to form their own conclusions...

Instead, John Stewart made a joke about validating his preconceived ideas concerning Republican responsibility.  He played a clip of a Republican member of the Defense Appropriations committe comparing AHLTA (the DoD healthcare information system) and VistA (the VA's Healthcare Information umbrella) to Play Station and X-Box which, though both can use the same TV, cannot talk to one another.  AND THEN Mr.Stewart suggested that the parent of the household, if he/she wanted to minimize contention and confusion, could impose a single solution on the household.  A photo of President Obama was displayed during this, clearly indicating that all the blame could be laid at the President's feet.

Nothing could be further from the truth.  The VA has been assembling the components of VistA since 1987  and it was fully formed by the late 1990s.  AHLTA was introduced amidst much fanfare at the beginning of 2004.  The birthing process for AHLTA overlapped the adolescence and maturation of VistA, indicating a conscious decision (pre-2004) by the DoD to ignore the VA's efforts.

This entire mess is too complex in its origins and effects to discuss in a single blog entry so I'll content myself with one last piece of exculpatory evidence on behalf of anyone who has assumed elected office since 2004.  The only way to move information from AHLTA to VistA is

1. Generate paper from AHLTA
2. Send the paper to the VA
3. Enter the information from the paper into VistA

This is also the reason why the DoD can't simply replace AHLTA with VistA. 

There is a lot more to be said on this subject but, among other things, it seems clear that this is but one (albeit very painful) example of the inability of our health care "system" to adopt anything approaching a standardized view of the enterprise.  Interoperability is something to be wished for but which cannot exist in an environment in which healthcare systems vendors are the X-Box, the Play Station, and Wii.

Data is like Scat

We (data people) like to talk about data as if it were the most important thing in the world when, in reality, it is exactly like turds.  Yes, that's what I said, merde, gavno, scheiss, shit. 

Of course, it could still be the most important thing is the world but whether it is or is not depends on many things, not the least of which is what we're trying to do.  Data is the byproduct of process much as turds are the byproduct of digestion.

For naturalists and scientists castings or anaimal feces provide important information on a wide range of subjects including the animals diet, where it has been, and its general health.  For a tracker, a different set of information is deducible from the spoor.  The tracker tells which species, how recent, whether the animal is moving or is likely to be nearby and other things that may help him earn a bonus.

No one says, "This is quality shit." or "We need a governance program to improve the quality of our shit."  They simply learn what they can from it and move on.  The kind of value available changes over time but even petrified feces have a story to tell.

If we presume to be data experts, maybe we should be focused on extracting value from our data and understanding the kinds of value that can be extracted as well as how to recognize the data that will tell us what we need to know.

Even the most inconsistent set of data has a story to tell and we should listen to that story instead of wailing about the story we wanted to hear.  Because data is the byproduct of process, inconsistent data should tell us that the process that produced it is inconsistent.  If this is the case, how can we expect consistent data unless we can create a consistent process.

The bottom line is that when we focus our efforts on data quality we are misunderstanding the world we live and work in.  We are creating additional and entirely unncessary complexity.  How does this happen?  The cause lies in the storage of data for input to and output from computer systems.  We have allowed ourselves to institutionalize the cart before the horse.  What computer systems require is consistent input.  A system can be designed to deal with quality issues as long as they are predictable.  A system (process) will always produce output of a consistency equivalent to its input.

If we were pursuing consistency instead of quality, our disagreements would be fewer, our stress would be lower and our impact would be greater.  Consistency a readily understood concept while quality will always be the most elusive of quarries.