[ Who was Rear Adm. Grace Hopper ? ]

My apologies to BYTE magazine, but their archives only go back to 1994. This article [ "What Lies Ahead" (Grace Hopper), 1989, Byte Magazine, January, 343-345] should be required reading for any IT student. I [D.B.Cooper] came across this at a time that I was working on an inventory control project. It put a lot of things in their proper perspective.

Prioritizing Information
Rear Adm. Grace Hopper, USN, Ret., speaks out on how we must analyze, evaluate, and use the various kinds of data that we input into computers.
Compiled by Janet J. Barron

Model Ts totally changed the world of transportation. They cost between $300 and $600, and all of a sudden, people could afford to buy and own cars. The whole world changed because of the Model T, and yet, somehow, we totally neglected the underlying concept-transportation as a whole. Because we didn't look at transportation as a whole, today roadbeds and railroads are falling apart all over the U.S.

Now we're at the real beginning of a relatively new industry [the computer industry] that will eventually be the largest in this country. And I am very much afraid we'll make the same mistake all over again. I'm afraid we'll continue to go out and buy pieces of hardware with flashing lights and lovely "user-un-friendly" software and totally neglect the underlying subject--the total flow of information through any organization, activity, company, and so forth. We should be looking at the information flow and then selecting the equipment to implement that information flow.

Of course, to use our best equipment for handling the most valuable information, the first thing we need to know is which is the most valuable information. For a couple of decades now, I've been asking people how they value their information. I haven't really received any answers, but I have received a really great assortment of blank stares. Some people even question that there's a difference in the value of information.

I know of an oil refinery that's operated by computer. Information comes in from marketing and goes to the computer, which opens valves and pushes stuff through pipes and tells inventory how much of the finished product has been made. The computer puts out payroll reports and makes out the checks, as well as generating reports on all the activity that occurs.

Let's suppose that two pieces of data simultaneously enter that flow. One comes from a valve out in the plant and says, "If you don't open me, the plant's going to blow up. You have 45 seconds to act to save 78 lives and a $120 million plant." At the very same instant, from another part of the system, comes the fact that Joe did 2 hours of overtime. Which is the more valuable piece of information? And what are our criteria?

Finding the value of data
Most large companies insure their databases against damage, inability to access them, and other perils. What happens when the insurance company or the FBI asks, "How much is it worth?" Probably one of the biggest jobs we have ahead of us is to determine that value.

We have totally failed to consider the criteria for the value of information. We haven't yet even defined our criteria. And yet we must know something about the value of the information and data we are processing.

I think we must create several priorities: the time you have to act on the data and the number of lives at stake. But there's another one--the importance of that piece of data in making decisions.

We have completely failed to analyze the raw material that we are processing. We've spent almost half a decade talking about the process, the system, and the training. We've forgotten to look at the data or at the information we are producing. We've also neglected to notice something about information: Information is totally inert; it never does anything. It's something you see on the printed page, or you may see it on a computer screen, or you may hear it over a telephone. Although it never does anything, it still has to be fed through another process.

We have a raw material that is called data. We feed it into a process. In this case, the process consists of hardware, software, communications, and trained people. Hopefully, the output product is information. Equally hopefully, this process is under some from of control and there's a feedback loop from the information to the control to improve the quality of the information.

This process consists of a human being who analyzes, correlates, relates that information and turns it into something we can call knowledge, or intelligence--that we can make decisions on. But since we've added another process, we have to have another means of control and another feedback loop to improve the quality of the knowledge.

Take the example of a manager of a service business. Say he has 150 computers in his company, all providing data to him. He tells you, as his office manager, that he wants to know how much the firm has spent on gasoline for its fleet of vehicles. You can tell him how much it has spent on gasoline, but that's an insufficient answer.

Noting how many months there are left in the fiscal year, you should tell him how many gallons of gasoline the fleet has used so far and how many miles each vehicle travels on the average per month. Considering the possibility of an oil shortage, you can also project how much each gallon of gasoline could end up costing and, with some what-if scenarios, give your manager a rundown, in both figures and charts, of several possibilities of what the firm may have to spend on gasoline for the rest of the fiscal year.

We have got to give people better answers--more complete answers. This is where we're going to need the little expert routines--not the great big ones, the little ones--and we're going to need them very badly. One of the biggest jobs for these applications will be to find out what people need to know. What does the manager of a business need to know? We've got a tremendous job ahead of us.

There's something else that is important about this information business. Information has an actual value to a corporation. Someday, under "Other Assets," there's going to be an entry on the corporate balance sheet--"Information"--with a value on it. I asked the IRS how they plan to depreciate the value of information. They didn't answer me.

A Question of Accuracy
There are some other issues regarding information that we need to address, such as accuracy--correctness of the information that comes from your computer. One year I turned in my budget, and even though it added up right and had beautiful charts in it, it was rejected for inaccuracy. I put it through my computer again, and it came up with the same answers. Certainly, it was correct because it came from the computers. We've gone that way for too long.

When one of my supervisors heard about this situation, he decided to find out the possible costs of incorrect information in a data-processing system. He found a section of the privacy law applying to government employees in the military that dealt with this issue.

It states that if there is inaccurate data in a a personnel file and, because of it, someone is denied a raise or a promotion or otherwise treated improperly, they have the right to sue the federal government. The fact this law is on the books shows how serious a situation the military considers the possible consequences of inaccurate information. There are few cases an individual is given the right to directly sue the federal government.

This officer decided to analyze the possible costs to the government of inaccurate information in a personnel file. He used a hypothetical case and extrapolated, using statistical probabilities, and estimated the damages in a best-case scenario.

He came up with the fact that in just one possible situation, inaccurate information in a few personnel files could cost an organization in the range of $500,000. Now, this half-a-million-dollar projection was just for one not-so-bad case of incorrect data unintentionally finding its way into a small percentage of one department's personnel files.

You can use your imagination and multiply that figure by a factor of n. Then you can see for yourself how many millions of dollars of damage (not to mention the grief of those who get turned down for jobs, raises in salary, and so forth) that inaccurate computer information can cause. And the issue can't be resolved by going to your boss and asking for hundreds of thousands of dollars to correct files and save the company money. We've got to find out something about, and analyze, the cost of incorrect information, and it's a subject we've totally failed to address.

The Flow of information
Consider the concept of a river. It starts with little tiny rills, tiny drops of water going downhill. These drops coalesce into a small brook. That process continues on and on until somebody decides to use some of that water. So, they build a little dam and create a small pool. They use the water there. Then it coalesces again and runs on down and finally gets to be a good-size brook. and the brook runs into another dam before it enters a lake.

People use the lake water locally before they send it into the lake itself, from which some of the water flows out into a main river, which goes into some part of the state. Everywhere along the way, there are dams and reservoirs, and the water's used, coalesced, and sent in a broader stream on down until it finally reaches the sea.

I think data ought to behave this way. It should be collected locally at a branch office of somewhere and used there. Then it ought to be coalesced and forwarded to a regional office and used there, coalesced, and finally end up at a headquarters. Then there also would be a reverse flow that matches the river flow.

Telephone lines parallel rivers. When headquarters decides that something needs to be done, it sends orders back out to all these parallel branches, out to the branch office. That's a real flow of information, where the data goes both ways, and it's used locally before it's sent to headquarters. It's not centralized. There's no sense in sending every detail that concerns the branch office into Washington, D.C.

It makes sense that when the data-processing and MIS departments within a company look only at the computers and the telephone lines that they control, they're not doing their jobs. They're not looking at the information itself, at the value of that information, or at who uses it and why. We have to learn to manage information and its flow. If we don't, it will all end up in turbulence.

Acknowledgment
The information in this article was from a speech by and an interview with Grace Hopper.
-------------------
Janet J. Barron was a BYTE technical editor at the time this article was printed.

Return to Executive Summary
Copyright 2001 Green Meadow Enterprises, Inc.
Last updated 09/05/2001