World Futures Institute: Modeling – Part One

Los Alamos World Futures Institute

In making decisions about what action to take we create models, mental or mathematical, to help make our decision.

The Cheshire Cat continues harassing us about knowing where we are going or it does not matter which road we take. If very few variables are involved and the quantities are small, it is easy. But as the numbers grow, disciplined modelling is appropriate.

Let us start by building a “silly” model for what most of us would consider a “silly” issue to investigate.

Assume the issue is time reporting for a company, institution, agency, or lab that employs exactly 10,000 people. It is important that every employee records his or her time on the job so that full value is received for salary paid (40 hours per week).

The average salary is $50 per hour for 2,000 hours of work per year and 10 days (two weeks) vacation. Hence, the average annual salary is $104,000 per year. But wait, we have not included sick leave, paid holidays, or the extra day (or two) per year. After all, 52 weeks is only 365 days. But let us use $50 per hour since it will not “make any difference”.

Each employee reports his or her time via a networked computer that is always in his or her use. Every day they must report arrival time and departure time if they want to be paid. Obviously, identification cards that trigger an electronic recording system could be used, but for this “silly” example it must be done from the employees’ computers.

When an employee’s computer boots up and goes online, timing ignored for now, the employee must find the appropriate button on the screen to click and then click on it. For the model being built, assume this takes one, two, three, or four seconds. On clicking the button, another screen appears requiring the employee to click either the “log-in” or “log-out” button. Making the selection takes another one, two, three, or four seconds.

The average total time for clicking is five seconds (you do the math). Of course, this time may vary, especially if the employee is interrupted during the process or otherwise is distracted. But for this model we agree to use the five second number.

If every employee does the log-in/log-out every day, that amounts to 50,000 seconds every day or 250,000 seconds each week for the entire workforce. Since employees only work 50 weeks per year, the time used comes to 12,500,000 seconds per year.

Dividing by 60 seconds per minute, this equals 208,333.3 minutes (to limited accuracy). Further dividing by 60 minutes per hour, this amounts to 3,472.22 hours which, at $50 per hour, comes to $173,661.11 in cost. And the range in cost, using the range of 2 to 8 seconds in the model, is $69,464.44 to $277,857.73.

With 10,000 employees at $104,000.00 per employee per year, the sum of annual salaries equals $1,040,000,000.00 of $1.04 billion per year and the worst-case time entry cost represents 0.0267170942 percent (notice the accuracy?).

Who cares? But in contrast, if it is publicized that the company, institution, agency, or lab is “wasting” over a quarter million dollars per year counting time and attendance, what is the reaction among the public and within the organization’s management?

At the beginning of this model it was “assumed” that all 10,000 employees had to record their time and attendance daily and that it would be recorded.

This was done without proof, simply being used for this model. Is it really true or is it just sort of true? Can one employee log in for others simply by starting their computers? Do the time elements have to be absolute? How do you handle lunch breaks? And why can you assume that the clicking can take only 1, 2, 3 or 4 one, seconds? Perhaps samples of clicking time need to be gathered. After all, we could be wasting $277,857.78 per year. A study at $125,000.00 cost is clearly worthwhile.

Building a good model for insight during decision making requires reliance on data. In the study above, assumed funded, “extremely accurate” model samples were taken by observing the function of the employees and measuring entry time. This resulted in a nice bell shapes curve (maybe) with an average value and a standard deviation.

Great data, but what was the collection error? Was the observed employee’s environment altered by the process? And did it affect the results? What data do you need, how good is it, and do you trust the source? And how good are you at assessing the data needs?

Til next time…

Los Alamos World Futures Institute website is Feedback, volunteers and donations (501.c.3) are welcome. Email or email Previously published columns can be found at or