Software engineer questions what future scientific innovation looks like, how productivity is diminishing and what can be done

By Clara Lincolnhol

This is the 1^st in a series of articles by Knight Center students who attended the recent annual conference of the National Association of Science Writers.

Is scientific innovation slowing down?

That’s what Cristian Ponce, the CEO of software development company Tetsuwan Scientific, asserts.

He also urges people to look into the effects that diminishing research productivity has on creating life-changing products like medications.

When William Shockley invented the transistor in 1947, that allowed the modern computer as we know it today to exist, Ponce said, speaking at the recent annual conference of the National Association of Science Writers in Chicago.

“Innovation in the transistor has enabled computers that are faster, more affordable, more energy efficient and much smaller in footprint,” he said.

There have been great jumps in technological innovation due to the development of transistors. These technological advancements coincide with many changes in society, he said.

“If we’re able to describe how the computer changes over time, we are able to describe in many ways how society changes over time due to computers’ important impact on our society,” Ponce told an audience of science journalists, writers and communicators.

In 1965, an engineer and business executive named Gordon Moore observed that the number of transistors in a circuit doubles around every two years at the same energy expenditure, leading to greater technological development.

Moore called that observation “Moore’s Law,” Ponce said.

“Every 18 months the computer doubles at the same energy expenditure for a computer, and this trend has remained true for a very long time,” he said.

But Ponce says Moore’s Law doesn’t accurately reflect what’s going on in the tech industry anymore. In reality, now it takes significantly more energy, time and resources to maintain that trend of innovation — about 18 times more resources than in the 1970s, he said.

Maintaining Moore’s Law is unsustainable, but companies still try to do so because of the importance of computing in many industries, he said.

The tech industry, he said, has “been willing to throw more and more and more capital and resources at this curve to make sure that it stays at the same pace, but, still, Moore’s Law has begun to wobble,” he said.

Ponce also said it is misguided to think that because the industry experienced exponential growth until this point it must do anything to maintain that pace.

“They say, in order to maintain that curve, what we have to do is not regulate and provide more general money to these projects,” he said. “This is a dangerous narrative that’s not rooted in how innovation truly works.”

Preventing regulation of the AI industry as one way to keep innovation going is dangerous, he said.

Diminishing research productivity has a negative impact on the development of pharmaceuticals, Ponce said, citing Eroom’s Law.

Eroom is Moore spelled backwards.

“Eroom’s Law is the opposite of Moore’s Law. If we think that computing gets exponentially better, we know that pharmaceuticals get exponentially worse. Drugs get exponentially more expensive.”

Even with technological innovation, it’s becoming increasingly harder to develop medications, he said.

There are a number of hypotheses for why that’s true, including the “better than the Beatles” phenomenon and the “low-hanging fruit” problem.

The “better than the Beatles” problem is the idea that new drugs must be significantly better than existing treatments. Oftentimes, already existing treatments are the result of major progress.

For example, say there’s a drug that has a negative side effect of making a patient nauseous, but it works well enough.

If the developers of the drug want to fix that, it would be incredibly expensive — and then the cost of the drug on the market would rise immensely but it’s only 20% more effective than it was before, he said.

The low-hanging fruit problem describes how clinical trials are becoming increasingly expensive because going after diseases that are harder to treat requires more expansive and costly clinical trials, he said.

“We see this pattern where the money you can make in drug discovery keeps going down and the amount to try to make that money keeps going up,” Ponce said.

The abandonment of models that better predict the success of a drug, like animal modeling that better represents human biology, also creates problems, he said.

For example, a new model developed where big companies are not doing their own biotech research and, instead, are contracting out research to tiny biotech companies known as contract research organizations.

“This very convoluted model again has evolved in response to this exponential pressure. The system has contorted itself to be able to keep the lights on.”

Large pharmaceutical companies like Johnson & Johnson and Pfizer used to do research and could survive economic duress, but now small companies are doing the research, he said.

But those small companies cannot survive hard times and are under the control of the large companies, he said.

“So, this model is more efficient, but it’s also much more prone to failure. It’s a lot more delicate, he said. “It’s also a system that has a lot of hidden exploitation.”

Ponce said the productivity trend is applicable not only to pharmaceuticals but also in fields like agriculture, and he said he expects it to worsen under the Trump administration.

One way to respond is by taking new risks, even though historically the National Science Foundation and National Institutes of Health are risk-adverse. They have usually funded only research endeavors where positive results and data are expected, Ponce said.

“This fixation on finding things that we know will work is only worse for us,” he said. “We have to take samples, we have to take risks, we have to try new things. Otherwise, we will fall into the same systems of diminishing productivity that we have fallen into in the past 70 years.”

He also said it’s important to restore, replenish and maintain communal resources that are needed for research productivity because they are being depleted.