Optimistic cost-benefit analysis for technologies

Studying the best-case scenario exposes bad ideas and counters negativity bias.

There’s a framework I use when analyzing a new technology that I call Optimistic Cost-Benefit (OCB) the basic idea is:

New or speculative technologies should be supported or dismissed on the basis of the most optimistic yet physically grounded assessment of their value.

It’s crucial to bake optimism into your analysis because new technologies are butterfly ideas and deserve a chance to fly. Many technologies fail, but some succeed in a dramatic fashion, so it’s better to fund 10 ideas and see one succeed than to fund none.

Optimism counters a natural skepticism people have towards new ideas. By and large, people and media are far too pessimistic about new developments. Government scientific funders reflect this by being far too risk adverse with their funding choices. On top of all this, investors seem unaware of the basic economics around some of the technologies they’re funding.

If you’re open about the assumptions you’re making, OCB can increase clarity and aid decision making. Optimistic assumptions often make the problem simpler by assuming away certain costs. This kind of analysis makes it easy to rule out bad ideas; if something doesn’t work under the best possible circumstances, it’s safe to dismiss.

So I’m going to walk through how I set aside my biases and assess the best possible case for new ideas.

Make a framework for the costs and benefits of a technology

Usually, the best way to do this is to imagine how the technology could be used in a product in the future. Perhaps your idea cures cancer, or generates energy, or repairs satellites.

Next, look at the major costs that go into creating this product. You should be thinking about raw materials costs, the capital and complexity of building a factory, and the cost of labor.

Ask yourself if your idea will make a substantial change to the overall cost of the final product. If you’re optimizing something that’s only 5% of the overall production cost the gains will be limited and you’ll want to think bigger.

Some of these costs are hard to estimate, so one trick I use for simple products is to assume that raw materials make up the bulk of the cost. This is an extremely optimistic assumption, but it’s perfectly consistent with our goals. It’s a very clear and simple way to model the costs and as long as you’re upfront about it, I think it’s fine to do.

Assuming away capital and labor costs is fine for a simple process, but the more complicated your idea the more important these costs are going to be. Computer chips for example are dominated by equipment and labor costs, the raw cost of silicon and energy is a tiny portion of the cost to make a chip.

For estimating capital and labor costs, you can try to look at similar processes and assume you can do it for roughly the same price. Perhaps you’re considering a fancy lab-on-a-chip, you can ballpark the cost by using existing semiconductor prices. This is once again an optimistic assumption, bespoke hardware will always be more expensive than mass produced hardware, but as long as you’re explicit about it that’s fine.

Flipping over to the benefits, ask what benefits the product has. How is it better than what we have today? Can it accomplish something faster or cheaper? Is it better along some metric?

If possible, try to estimate the profit you would get from the new technology. If you can make a kg of plastic for $2 and the current price is $10 that could be pretty valuable!

More often than not, though, it’s pretty hard to figure out what people will pay for a new technology. If a new mRNA cancer vaccine extends survival times by 15 months, how much profit will that generate? Instead, just try to estimate the benefit and compare it to existing solutions rather than trying to price in these benefits. This makes it easy to compare to existing technologies.

Consider how things might change once a technology like yours becomes commonplace. As gene sequencing costs have fallen, biologists have started using DNA ticker tape to record events in the cell, since reading this DNA information is cheap now. Some of the greatest innovations have reshaped an industry by suddenly making something cheap. The new efficiencies your idea creates should count as a benefit.

Model costs and benefits with explicit formulas

The key here isn’t to get an exact mathematical description of the economics, but to make your arguments concrete and easy to analyze. Your analysis should be clear to others and make it easy to see the load-bearing assumptions.

Because the formulas are mainly for communication, it’s better to keep things simple at the expense of accuracy. Be explicit about the simplifications you’re making and why they are reasonable. As always, the simplifications should paint the idea in the best possible light while being physically realistic.

Here’s an example: in Rocketplanes I needed a way to estimate the trajectory of a rocket flying to different parts of the world. The real formula for this trajectory is pretty complicated, but the absolute shortest distance a rocket can travel is along the surface of the Earth (i.e. the great circle distance). The true trajectory won’t be too different, but this is still an optimistic assumption.

Once you have some formulas for the costs and benefits, try plotting different things against each other. How sensitive are your conclusions to small parameter changes? Is there an optimum along different axes? Do you see increasing or diminishing returns? These charts will inform your thinking a lot.

Make a fair comparison to competing ideas

Formula in hand, you can plug in numbers to get an estimate of how good the idea is. If the main factors of your model aren’t numbers you can easily look up, consider refining it. You can always estimate the inputs, I’m a fan of rounding estimates to the nearest order-of-magnitude since it makes calculations clearer.

Now it’s time to see how your idea stacks up against alternatives. What’s the current price for a similar product? Is your idea cheaper in theory? Is it better in some other way?

For a fair comparison, you have to make an optimistic assessment of competitors’ future prices. Look at where the industry is going, will they be much cheaper in a few years? Are there improvements that will make competing products better? It takes a long time to commercialize something, so your idea will be competing with future technologies, not today’s technologies. That being said, you can mostly stick with today’s prices if an industry has stagnated.

After all that, we’re looking for order of magnitude improvements over the state of the art. Since we made so many optimistic assumptions, the OCB estimate is an upper bound on how good the idea is. If it’s not way way better on paper than what we have today, there’s a good chance that it won’t work in practice.

What counts as valuable depends on the field. If you’re proposing a simple modification to an existing process, a 10% improvement might work out. If you’ve got a complicated fusion reactor design, it needs to be much better in theory to justify all the risks of implementation.

Conclusion

An OCB is a great way to wrap your head around a new technology and it naturally biases in favor of new ideas while giving you tools to think about the bottlenecks. Once you’ve got a good model and have thought through the different issues, you can put it in front of an expert. They will bring up details and ideas that you never considered, which can deepen your analysis.