Feedback

TechWatch: Connecting With Digital Technology

Few would want to step back from the digital age. Digital technology has brought to us all sorts of conveniences and efficiencies for home and work. Yet it is remarkable how little progress we have made in understanding how to interact with this technology, whether it is the “netiquette” in using it or the meaning associated with the data we generate.

Cell Phones in Africa

This came home to me on a recent trip to the Central African Republic. This is one of the world’s poorest countries, and when I first went there two years ago, there was little evidence of a digital age. Last year there were a few cell phones. This year cell phones were everywhere, and the most successful vendors on the streets were selling phone cards good for a few minutes of time.

As with the rest of the digital revolution, this technology provides wonderful opportunity for a poor country — it can leap past the expensive wired infrastructure and begin gaining the communications benefits at a small fraction of the entry fee. But this also has its downsides. Cell phones would go off in the middle of high-level meetings, and those with the phones would answer them without a thought. When my translator’s phone rang in the middle of a seminar, he took the call and somehow continued to translate! I hope over time they will begin to determine a better way to interact with the digital age.

Lest we believe we have this figured out, however, I have observed that we still struggle with the data that is generated from this digital system. I offer a few illustrations, from small and fairly obvious to the world of managerial accounting.

Digital Precision

Before the digital clock, we were pretty good with providing an appropriate accuracy for the time. We might say it is about half past the hour, when that is appropriate. Or we might say it is 8:27 if the next train is due to leave at 8:29. But with a digital clock, someone will say it is 8:47, even if the clock is running a couple of minutes fast and the context requires no such precision. The precise answer is not always warranted by the data or the situation.

It reminds me of the person flying over the Grand Canyon, telling the person next to him that the Grand Canyon was 2,000,000,002 years old. When asked how he knew this, he said he was flying over it two years before, and was told it was 2 billion years old!

While these examples perhaps seem only mildly amusing, they work their way out in a range of interesting ways. The clerk at McDonald’s offers you $12.32 in change for a $10 . They obviously put in a wrong amount somewhere in the process, but the transaction never passed through their head. Or the manager asks a support person for the sales forecast for the next month, and gets the response: $15,247.63. This may be the number computed from forecast inputs and probabilities, but again the precision is not warranted. Perhaps an unquestioning belief in what comes out of the computer is the problem here. We may have forgotten that “garbage in, garbage out” still applies, and imprecise data does not yield precise results.

Sometimes the person creating the data is at fault. I remember earlier in the computer age looking at a computer run (from the batch computing era). At the end of the printout there was a table:

CPU used: 495.2 seconds
Storage used: 52,486 bytes
Tapes mounted: 2
Pages printed: 27
Total resources used: 53,010.2

This is, of course, precisely correct and completely useless.

Understanding and using digital data requires more thought than reading the output. It can also get us into trouble in another way. Because computation is so fast and cheap, we compute things that seem like a very good idea at the time, until we understand the downstream implications.

Managerial Accounting

This takes me to the world that I now know of as managerial accounting. One definition of the term is this: “The branch of accounting that uses both historical and estimated data in providing information that management uses in conducting daily operations.” We have lots of data available, and can perform lots of calculations with it at little cost. So it would seem that this area could offer all sorts of benefit to any business. And it can, but …

Here is an illustration of managerial accounting going off the track. It is an older example from my personal experience more than a decade ago, but I expect it will relate to similar issues in the business world today.

The Boeing Company had purchased a $15 million Cray supercomputer, capable of doing the type of intense calculations needed for the design of the next-generation airplane. But the finance people determined that such a valuable and costly resource should not be put into overhead, but should be charged back to the users. They created a very complicated algorithm involving some combination of the resources used. This seemed like the right thing to do, and usage grew on the computer to almost half of full utilization, until the next budget tightening exercise hit the company.

Departments such as aerodynamics research that normally used this computer suddenly realized that they could save money in their department budgets by moving their computer programs to work stations, since they had no accounting charges associated with them. Yes, they would now run slower — they might take 24 to 48 hours to compute something that would be done in less than an hour on the supercomputer. But in these days of tight budgets, they would simply have to wait. So high-priced scientists spent weeks of their time moving computer codes from the supercomputer to the work station in order to save money for their department. This meant the company was getting far less production from these same people (many fewer computations, much time in moving the codes) at no savings to the company, since it still owned the supercomputer.

There was a compounding effect. When the first department moved its codes, everybody else who used the supercomputer saw their bill for use go up. That’s because the total cost of the supercomputer center was allocated to the users each week, and with fewer users, each usage became more expensive. The use on the system dropped to less than 15 percent. It couldn’t drop any lower, because there were some things absolutely needed in the company that only the supercomputer could do.

Several of us came together to try to address the issue of the underutilized supercomputer. One idea was simply to remove it. But the second idea was to address the managerial accounting solution that had driven this behavior. By simply putting the computer into overhead the problem would immediately go away. This would mean that anyone could use it for “free,” and there was a concern that important work would be crowded out by less important work. So we settled on another managerial accounting solution: Charge the use to the users as if the computer were fully utilized. No one would be penalized for a light load some week. The modest charge might be enough to filter out unneeded work.

Once everyone had agreed on the final solution, I met with the two more senior executives who were a part of the solution development, and one asked how long I thought it would take to start fully utilizing the supercomputer. We decided to have a small wager, no money involved. One executive put in two years. The other put in one year. I put in two months. We were all wrong. The computer was fully loaded in one month.

Keeping It Simple

Realizing the potential for unintended consequences should be a reminder to keep it simple. Complexity is possible for the models of cost allocation using our modern computers. But the complexity has its costs. I once thought that theatre’s and stadiums should take advantage of today’s computing capability by eliminating the one-priced seating sections. It would be fairly easy to assign a value to each individual seat, eliminating the problem of having a $25 seat in the row behind a $40 seat. But with individually priced tickets, the complexity would go up for each person buying a ticket. I am not sure it is worth it.

Lessons Learned

The lessons should be obvious. Digital computers make it possible to perform very rapid calculations with complex data. Managerial accounting is a tool to assure effective use of resources by allocating costs “properly.” Be careful. The results of the analysis may drive surprising and unexpected behavior.

This does not mean a business should not use data, made possible by the digital age, to drive its business. It does mean models for such systems should be kept as simple as possible, and there should be a period of analysis of potential unintended consequences before implementing them.

Just because you can compute something doesn’t mean you should.

erisman-thumb

Al Erisman is executive editor of Ethix, which he co-founded in 1998.
He spent 32 years at The Boeing Company, the last 11 as director of technology.
He was selected as a senior technical fellow of The Boeing Company in 1990,
and received his Ph.D. in applied mathematics from Iowa State University.

Share Your Thoughts