Marc Andreessen's Turkey-Optimist Manifesto
The Techno-Optimist Manifesto Mistakes How Technology Operates, and Misses the Problem of Induction
Marc Andreessen reasons like a turkey.
The Techno-Optimist Manifesto, by Marc Andreessen, argues against the AI doom narrative, and makes a strong case for technology making the world a better place.
But he forgets what the stuffed, dead turkey taught us.
In his Techno-Optimist Manifesto, Andreessen writes:
We believe that there is no material problem – whether created by nature or by technology – that cannot be solved with more technology.
We had a problem of starvation, so we invented the Green Revolution….
We had a problem of cold, so we invented indoor heating….
We had a problem of isolation, so we invented the Internet….
We have a problem of poverty, so we invent technology to create abundance.
Give us a real world problem, and we can invent technology that will solve it.
He is using past inventions and successes, and predicting that the future will resemble the past.
This is using induction - taking from what we know - and using it to predict the unknown.
David Hume argued against induction, claiming that:
Let the course of things be allowed hitherto ever so regular; that alone, without some new argument or inference, proves not that, for the future, it will continue so.
And, as Hume proceeded to claim:
That the sun will not rise tomorrow is no less intelligible a proposition, and implies no more contradiction than the affirmation, that it will rise.
Or, in modern English: Because we know that something happened, it does not prove it will continue to happen; past performance does not guarantee future results. The fact that the sun rose is not a reason that it will rise; the past cannot show us the future.
How, then, can we know anything that we do not yet know?
David Deutsch, philosopher of science, explains that induction does not work by counting cases. Rather, it works by finding the structure that connects them, and understanding what makes them work the way that they do.
Induction does not work by counting cases.
Rather, it works by finding the structure that connects them, and understanding what makes them work the way that they do.
Copper conducting electricity is an example of good induction.
We know that copper conducts electricity, and we are certain that copper which we have not yet seen will also conduct electricity. We are sure, not because we saw that it has conducted electricity, but because we know that copper has one mechanism which conducts electricity.
Because we understand the why, we can know even what we have not yet seen.
This is an example of good induction.
An example of bad induction is the turkey problem
The turkey is fed every day for months before Thanksgiving. It thinks that it will be fed forever, as it always was; it uses the past to predict the future, without understanding why it was fed.
Then, on Thanksgiving eve, the turkey is slaughtered.
The turkey’s mistake was not that it used induction. The mistake was that it used induction without reason; that it did not have a reason, or theory, as to why it was fed; it counted cases and assumed the future will look exactly like the past.
Andreessen is reasoning like that turkey.
He is using past technologies to predict future ones, without discussing the reasons behind it.
He misses that technology has no shared mechanism between each other. Each invention is completely independent, and not comparable to others.
The printing press, penicillin, and AI share only a label, but not a structure. Each one is its own object, working on its own physics, and is fighting against its own problem. Nothing runs underneath them that made the last one work.
The list that Andreessen uses is not one mechanism repeating itself. Rather, it is a set of unrelated wins, all lined up to create the illusion of a pattern; it is a set of wins that happened to have worked out, and do not prove how future technology will look.
Many push back on the manifesto, claiming that there are bad technologies created, that some do not help, or that there are bad incentives in the market.
But in truth, the fallacy of the manifesto runs way deeper. Past results do not guarantee future performance, and previous innovation does not prove anything about future innovation.
…the fallacy of the manifesto begins way earlier.
Past results do not guarantee future performance, and previous innovation does not prove anything about future innovation.
-Ralph Burton
That the manifesto is wrong does not make me pessimistic.
I am optimistic about technology, and how it can help the world.
But, as Andreessen himself writes:
We believe in the words of David Deutsch: “We have a duty to be optimistic. Because the future is open, not predetermined and therefore cannot just be accepted: we are all responsible for what it holds.
Thus it is our duty to fight for a better world.”
Deutsch is not a turkey, as Andreessen mistakenly thought.
Deutsch knew that the future is open. We should be optimistic, and it is our duty to make the world.
It is our duty to protect against failure, to guard against danger, and to create a better future.
Strongly worded manifestos urging the public not to care is reckless and irresponsible. It is not fighting for a better world, but is laying back and expecting the world to come to you.
Cautious, mindful discussion is what will help. Thinking, without the dogma of “we believe”, is what will improve our future.
We are better than that turkey.
We can plan, think, and take precautions.
We can reason and know more than just counting history.
We should be optimists.
And we should work on making the world a better place.
Which is exactly why we must be careful.
About Tech Without Myths
Tech Without Myths is a newsletter devoted to mechanism-first thinking about technology, the companies that create it, and the people created by it.
Atomic essays, published irregularly, on whichever part of tech is consuming my mind.
Beautiful argument for Optimism