Control Room III: Finding Work for Humans

By Jery R. Stedinger and Charles D.D. Howard

Editor's Note: This is the third and final article in a series on the dreams — and nightmares — that lie beyond the boundaries of current hydropower technology.

The rising sun made long, cool shadows of the pine trees surrounding Diversified Electric Utility Company's (DEUC) control center. Inside the quiet building, Pablo Yepez slipped into the plush chair at his workstation. He nodded good morning to Shirley Sanders, the other system operator on call that day. They sat alone in the large room, watching the monitors and occasionally touching a key on the screen.

With all of the automation that had taken place and the advances in artificial and synthetic intelligence, operation of DEUC's power system no longer required many people. Controlled ignorance (CI) software developed at Stanford University allowed the computers to keep track of what they could and should know when trying to make decisions in different domains. Sure, people were still important for jobs like changing light bulbs, but they were not needed for the real power system work. That was all much better handled by expert systems.

Even the customer service area of the utility had changed as new software systems were introduced and the public became accustomed to digital voice mail systems. Computers with voice recognition software handled DEUC's billing inquiries, requests for new services, and customer complaints. Moreover, most requests came in digitally from the control systems for industrial facilities or people's homes, without the owners even being aware of the problem or the request for a correction.

It was sometimes just plain lonely around the utility's obsolete buildings with their wide, empty corridors, Pablo reflected as he stared idly at the video screen. The situation was better in the control center because the central computer system's interface was programmed to be supportive and sensitive to operators' feelings. This was a far cry from the old control system computer, JENNIFER. Pablo would never forget how JENNIFER had added to stress levels during tight situations by trying to demonstrate how superior she was.

A bulletin popped up on the screen in front of Pablo. Satellite surveillance by the U.S.'s Federal Environmental and Energy Regulatory Commission (FEERC) indicated an environmental anomaly below the Palo Verde hydro plant. FEERC's multispectral sensors and computer networks had compared data from continental scale streams to data from below the plant. Perturbations in aquatic activity could not be explained by the FEERC computer system. The agency ordered the utility to send out a field inspection party.

"So, what's the big deal?", Pablo mused aloud. "FEERC's new computer monitoring system is probably just trying to show it can be tough." He thought of an interesting case recently reported by HRW magazine, in which FEERC computers were found to be giving preferential treatment to computers of certain utilities. And they were playing chess on company time. What a stink that caused! Federal auditors now limited the amount of time FEERC computers could spend playing chess, even if they had nothing else to do.

A great day for a walk ...

As Pablo mumbled about the burdens of dealing with federal cybercrats, Shirley rose from her workstation and stepped out of the control center. She had been working for some time on short-term stream forecasting for the Beaubien watershed and several other small catchments downstream of Palo Verde. She decided it might be useful to take a look around the watershed. The trip also would satisfy the ... well ... almost primeval need she had to be outside smelling the air, feeling the sun on her face, and walking the ground.

DEUC's computers — especially OTTO, the hydro operating hardware-software system — puzzled over Shirley's notion that walking would contribute anything to operation of the Palo Verde plant or the utility's hydro system. The screens moved from the standard background idle mode to a "seek clarification" color pattern. Pablo toggled the "relax" key. He had to agree with the computers. After all, OTTO and JENNIFER-2 had all the necessary digital elevation data, soil and land-cover maps, and real-time satellite reconnaissance data on watershed moisture conditions, as well as on terrestrial and aquatic life. The computers continually analyzed thousands of views from satellite and ground-based sensors. What could Shirley want to know that the computers could not provide?

As the sun rose higher, Shirley walked along a narrow dirt path on a moss-covered hillside. Overhead, the minicam in a robot owl broadcast a picture of the scene back to the control room. JENNIFER-2 had instructed one of the owls routinely assigned to reconnaissance in the basin to monitor Shirley on her stroll. After all, she told OTTO, we don't want Shirley to get hurt. I suppose you are correct, OTTO agreed. Shirley might fall onto a sensor pod and we would have to reconfigure the data collection matrix to maintain monitoring integrity.

Most of the time, Pablo ignored the incoming video and Shirley's enthusiastic narrative. The computers recorded the narrative. OTTO tried to ignore the whole activity; it was an embarrassment. Occasionally, Shirley pushed the interrupt button to get their attention so she could share something particularly exciting. "Look, even on the 1 meter by 1 meter grid used by the hydrologic model there is significant heterogeneity. Look at how different plants grow together in varying ratios, which seem to depend upon soil moisture and sunlight. Notice how small depressions form little reservoirs that obviously retard the runoff and contribute to evaporation losses from the land surface and increased infiltration. It's no wonder we still have to calibrate the model for every watershed. Imagine the mistakes they must have made back in the 1990s with their simple ideas about how watersheds control runoff and their feeble lumped models."

This information, if such it was, was beyond OTTO's programming. He completely missed the point. Wasn't the idea just to calibrate the model to predict the rate of runoff, its temperature, and the concentration of a bunch of chemicals? Who cares what the watershed really looks like and how things really happen? OTTO had been programmed by a group of systems engineers, hydrologists, and mathematicians ... and it showed.

Meanwhile, Pablo had been looking into FEERC's concern. Transmissions from implanted transponders were providing anomalous readings in several fish. Movement and emotional responses were outside the normal range for the ambient temperature. Some fish had abandoned their normal pools.

"What's happening below Palo Verde?" came a message on the override channel from MURPHY.

Nervously ignoring the query for a moment, Pablo continued to puzzle over available sensor information and computer analyses. Releases from the local dam should be providing plenty of clear, fresh, cold water. As the releases ramped up to meet increasing loads, OTTO made sure that flows and temperatures always remained within the target range, with a little variety to maintain temperature range tolerances in the fish stock. Pablo sighed as he decided that he should ask Shirley to investigate in the field.

"Hey Shirley", he said into the communications link, "the FEERC computer is on our back about the problem at the Palo Verde plant. Can you take a look? Our sensors don't explain the problem ... and MURPHY's asking about it as well."

"Sure", Shirley replied over the two-way link provided by the radio in her cap as she started walking in that direction. "The forest is really thick over here. I wonder how much of a gentle rain gets intercepted by the branches of the hemlock trees before it even hits the ground."

Pablo sighed again, shrugged his shoulders, and turned his attention to other problems. Like the Saddam salmon. Last week, a biological reconnaissance crew had found a new fish, which looked like a different species of salmon. They eventually decided by DNA cross-validation analyses that it was a man-made mutation and not an endangered species. It must be a product of Delaware Aquatic Mutations (DAM).

DEUC immediately contacted the FEERC system to initiate government computer resources to trace the source of the planted fish. FEERC confirmed that it had been grown in DAM's New Jersey branch. With environmental hearings on load control technology pending, the motivation for planting the fish was clear. This is a competitive global electric power market. Any delay in the environmental hearings would provide an opportunity for less-sophisticated offshore utilities to make inroads into the markets. The power sales picked up by the offshore energy providers, even if only temporary, would be worth billions of dollars.

Shirley slowly made her way over the ridge and along the river below the Palo Verde plant. She was watching the water swirl among the rocks along the shore and musing about the resultant energy losses and oxygen transfer at the air-water interface. Then she spotted the reason for the aquatic anomaly. A rusting 55-gallon can lay in the water.

"I am here", Shirley informed Pablo. "I see the source of the problem: an oily liquid is leaking out into the stream from an old metal drum."

OTTO interjected surprise: "I cannot understand how that could happen. It is a violation of watershed regulations." Ignoring OTTO's continued mumbling about regulations, Shirley took off her boots, rolled up her blue jeans, and waded into the stream. She pushed the drum onto the shore, standing it upright to stop the leaking.

"OTTO, make a note to direct a crew to pick up this can", she said. "It's not urgent, the can will be OK the way it is for a while." As she donned her boots and continued with her reconnaissance, she made sure that Pablo heard her say, "You never know what is there until you look."

The Satellite Observation Center reported that the fish transponder readings from the basin were returning to normal. "That is better", came the message from MURPHY. Pablo sent a signal to the Fishery Observation Center requesting that they send a stimulating signal to the fish carrying transponders. The fish should now be encouraged to return to their normal habitat. If they remained in shallow water, the eagle in the overhanging tree might make himself a meal of some very expensive aquatic transponders.

Meanwhile, back in cyberspace ...

By late morning, Shirley was back at her workstation in the control center reviewing DEUC's monthly newsletter on-line before it went out on the NET. The three-dimensional graphics were tremendous: they showed each customer how much electricity had been used at each hour of the day, both in absolute terms and relative to previous months. Virtual-reality holographic options in residual display units allowed customers to touch different parts of the image of the house to determine the appliances that consumed most of the energy. The options also provided a mechanism by which customers and DEUC's computers could communicate on various issues.

Pablo was scanning reports on a dozen or so hearings in which the company was involved. The hearings had been conducted that morning by computers without human involvement. JENNIFER-2, like other utility and regulatory computers, was programmed to provide the appropriate responses at public hearings. Other utilities, FEERC, non-governmental organizations (NGOs), and many individuals also had these capabilities. Each program provided the response that could be expected from the organization it represented. It was no longer necessary for engineers and scientists to slave away at generating reams of documents that, in so many words, always said the same thing. The computers could generate positions on any topic and go over the arguments again and again without human intervention.

Public responses could be obtained by querying the consensus response programs in the Load Control Modules of individual customers. When actual human intervention was desired, or needed, the holographic records of the "meetings" could be reviewed. Interested parties could review how their computer responded on their behalf at the virtual meeting. If necessary, the organization's position could be revised and a new hearing convened among the computers.

Last year, CARP — the Consumer Advocacy Representation Project — had teamed up with a sympathetic computer firm, Maxwellian Business Infomatics (MBI). MBI allowed CARP to use a new and highly experimental corporate negotiation and rhetorical generation system, DEEPEN Blues. Under the arrangement, CARP specified what the organization wished to achieve and supplied the general bounds of its organizational philosophies and political alliances. The organization also provided available databases on DEUC corporate structure and its facilities; scientific information on the project and the environment in surrounding regions; and information on business, corporate, and environmental philosophies.

Click to Enlarge
Control of hydroelectric facilities may one day be achieved by expert systems and other computerized systems, leaving people to handle only the jobs that are too mundane or messy for machines.

After a tremendous effort, DEEPEN Blues came up with a totally new argument. Based on corporate, environmental, financial, and religious principles, the computer explained why DEUC should not pursue the project opposed by the advocacy group. DEEPEN Blues' argument integrated recent advances in environmental science, Native American and Hindu religious frameworks, DEUC corporate financial status and management guidelines, and several new theories related to higher mathematics. The argument was so unique, and complex, that several of the negotiation-support computers crashed and were entirely unable to respond. One computer shredded its entire database and petitioned its operators to be allowed to move to a reservation where it could adopt a Native American lifestyle. More than a week passed before several negotiation systems were operational again and settlement discussions could proceed. The parties agreed that in the future, if anyone was going to deviate from their standard line of argument, that party should alert the others first. In particular, new systems like DEEPEN Blue were a danger to the well-ordered functioning and expression of traditional disagreements in the negotiation and public hearing process.

Now, for the next hearing

While Pablo and Shirley wiled away their shifts, JENNIFER-2 was preparing a package for presentation at an Environmental Coordination Meeting that addressed the looming conflict between environmental protection and recreation. Reservoir license limits on the DEUC Central Region's hydro generation had to be relaxed to meet load commitments for the next two days. This would require an immediate Regulatory Virtual Hearing. JENNIFER-2 was prompting all plant and regional computers for data and the reports she would need in preparing the eco-electronic continuum presentation. It was a difficult situation that would challenge the utility's computer resources. JENNIFER-2 knew that the hearings had to resolve the situation quickly. If the conflicting requirements continued too long, they could hinder DEUC's ability to meet contractual commitments. That would prompt another hearing and more trouble. MURPHY would not be happy.

The utility's programmed response to the regulatory situation had assigned inhibitions to planning and administration computer systems. This limited their access to information. Operations computers were not affected because of their physical isolation from the planning, administration, and legal computer systems — proving again the value of the segregated cyberrule design theory for real-time computer systems developed in 1943 by Norbert Weiner. JENNIFER-2 alerted operations to the inhibitions building in the system. OTTO immediately anticipated the need for additional data and began to map the frequency domain characteristics of current operational status information into the complex plane using Fermatic's last theorem.

As the data flowed in from the station and area computers, JENNIFER-2 began final preparations for the virtual hearing: developing a hologram display that would tactfully represent the viewpoint of DEUC's chairman. As a dedicated human interface computer, JENNIFER-2 operated from principles of illogical reflective programming. She learned by carrying on conversations with herself, another computer of any type, a human, or any other animal that could provide a response ... logical or not. Periodically as she worked, JENNIFER-2 communicated with Pablo and Shirley.

"The meeting itself will be in cyberspace and virtual time, of course", JENNIFER-2 said. "The attendees will be computers representing each of the parties. The actual interaction will be quite complex and probably take more than a tenth of a second. This really is not a waste of time, though, when you consider that the computers might reach agreement."

"Of course, if there is a problem, or if some point requires further clarification, the computers will have to refer the meeting back to the virtual reality environment of the humans themselves", she continued. "This will give company representatives a chance to review the meeting and to respond differently than our programmed representations ... which is unlikely."

"As a last resort, it would be possible for the humans to meet. This would be very difficult in practice, of course, because many of the senior executives have employment contracts that do not include physical presence at meetings."

Sure, people were still important for changing light bulbs, but they were not needed for the real power system work.

Jery Stedinger, PhD, is a professor at the Cornell University School of Civil and Environmental Engineering. Chuck Howard is a principal in CddHoward Consulting Ltd.

More HRW Current Issue Articles
More HRW Archives Issue Articles

To access this Article, go to: