Caleb Ontiveros

On Disproving Your Own Ideas

Disproving your own ideas is one of the most valuable things you can do.

If you don't do this you'll lack clarity, take on too much risk, and spin your wheels on the wrong projects.

If you want to think clearly, reduce risk, and work on the right things, you'll need to derisk your work. That will mean proving yourself wrong over and over again.

Suppose you're going through a todolist for a critical project. Whether it's writing, marketing, or building a company, there's an omnipresent question: what do I do first?

One rule you can adopt is do the hardest things first. The problem with this is that the hardest thing is not always the best thing to work on. Plus, there's the risk that you'll put hours into working on a hard problem -- with nothing to show for it a week later.

Another strategy take on the easiest tasks first. This way you'll be sure to get things done. This strategy builds momentum as you create success spirals. However, the easiest things aren't necessarily the best things to work on. Plus, many worthwhile projects just aren't that easy. By focusing on what's easy you may miss neglected opportunities.

Yet another strategy is do the most important things first. But how do you determine whether some task is more important than one another?

Think about this is in terms of derisking your projects, and then executing. If you derisk something, you'll have high confidence that it will work. Suppose you're working on building a company whose core offering is relaxing audio poetry. Here's an important question that you'll want to look into upfront: will people pay to listen to poetry? Is poetry like meditations (that people will pay for) or would people prefer to read poetry? Another question: will people pay for this offering instead of paying for a poetry book over audio? A common failure of first time entrepreneurs is to build an elaborate app. And then find out that, once it's built, no one will actually pay for it. Derisk!

Another example, when I was teaching philosophy, occasionally students would write entire papers on a topic before checking in with me about whether the thesis was  relevant and whether they understood the material. Before writing a paper, you'll want to ensure that you understand the research you are building on.

Whatever the cause, whether the task is too hard, the approach is wrong, or your thought is unclear, you don't want to work on what will fail. And if something is going to fail, you want to find out fast. Jacob Steinhardt formalizes this as a stochastic decision process. The key idea:

De-risk all components (to the extent feasible), then execute.

What does this have to do with disproving your own ideas?

Whether you're thinking about what to work on, constructing plans, or searching for intellectual insight, apply the framework of derisk and execute. In order to derisk well, you need to be good at disproving your own ideas.

Sometimes this looks like building an MVP or talking to potential customers. Sometimes it looks like setting a timer and writing all the reasons your thesis could be wrong. Sometimes it looks like the Stoic practice praemeditatio malorum: imagining all they ways your plan could go awry and realizing that you aren't prepared.

Evolution works by mutation, reproduction, and selection. By analogy, you need to select for the best ideas in your intellectual environment. To do that, ensure that the weak ideas don't survive. Disprove them. You can extend this analogy to idea generation, but let's save that for later.

That's the argument, here are three ways you can be better at disproving your own ideas.

First, be familiar with common techniques for disproving your ideas. If you're doing something where you can build a MVP, do that. Build the cheapest easiest version of the result and do that first. Sometimes this doesn't require building anything. If you're doing research, find a counter-example to your thesis. Refine it or move on. Try breaking down whatever you're working on to smaller pieces.

Second, become better at representing other points of view. In computer science there's this idea called the Turing Test. It's a test, originally constructed as a conversation,  that an AI needs to pass by being indistinguishable from a human. An AI would pass a Turing Test if you couldn't tell whether you're talking to a computer or a human for example. Can you pass Idealogical Turing tests for opposing points of view? If not, that's a risk.

Third, create an environment where your ideas will be disproving. Consume different sources of information. More importantly, put yourself in the situation where people around you will tell you when you're wrong. Publish your thoughts. Share them with your friends. Chose friends who will tell you when you're wrong. Thank them when they do.

When something fails, you want to know why. Knowing why is crucial. You can't lose, Recalling Epictetus' maxim:

If you seek Truth, you will not seek to gain a victory by every possible means; and when you have found Truth, you need not fear being defeated.

After you've completed derisking, move on to execution. Some communities and people spend too much time arguing that such and such is false or won't work. At some point, you need to stop thinking and start doing. If you're thinking well and have clear goals, the value of action will outweigh the value of seeking new information.

Against the Ontological Relevance of Meditation

You can’t learn about deep truths about the world from mindfulness meditation.

More precisely, you can’t come to learn any propositional knowledge about the world. This isn’t to say that mindfulness meditation isn’t valuable, it is. It’s a tool for increasing focus and discipline. And a way to practice reducing cognitive fusion.

Yet several claim that you can learn deep truths about the world from meditation. The fancy way to put this is that meditation is ontologically relevant.

We have a sense of being a subject located somewhere behind our heads, occupying the center of consciousness. We have bodies, but are not identical to them. Likewise, we have experiences, but are not identical to those experiences. This subject persists through time, has thoughts, and makes decisions.

Modern Buddhists, and those influenced by Buddhism, suggest that this sense of self is an illusion. On this view, there is no subject that:

  • Persists through time
  • Has a body and experiences
  • Has thoughts and makes decisions

Robert Wright calls this self, the CEO self in Why Buddhism Is True:

Buddhist thought and modern psychology converge on this point: in human life as it’s ordinarily lived, there is no one self, no conscious CEO, that runs the show.

Leaving aside the issue about whether this is true, I’m skeptical that you can learn this from meditation alone.

Here’s what you’re doing when you sit down to meditate:

  • Focusing on a single type of raw sensations like the breath or tingling in the body
  • Receiving all types of sensations

This dichotomy corresponds to training attention, the ability to focus on a specific thing, and training awareness, being conscious in general. To get a better handle on this distinction consider the visual field: there’s a point where you focus, this is attention, and there’s the entire visual field, including your peripheral vision, this is awareness. Think of attention as the focal point of the mental field, awareness as everything that falls within the mental field. Both practices can be combined as you train both.

As an aside, people often talk about flow as if it were one state. Flow is the wonderful feeling of ‘being in the zone.’ Really, there are two states here, there’s a focused flow, when you’ve zoomed into a single task, ignoring everything else. You’re completely absorbed in the activity. But then there’s also panoramic flow, in this state you’re in the zone, but the zone is much larger. Both of these states have their uses and risks. One can move into both states while meditating.

When you’re training your attention and awareness, you’re doing so in a nonconceptual way. Discursive thought, verbal chatter, these are at first distractions. One notes them and returns to the meditation object. Later thought becomes a meditation object itself.

This nonconceptual nature is easier to understand in the visual case. Suppose you’re meditating on what is front of you. You take in your visual field, not as a space full of chairs, tables, a water bottle, windows or whatever else you happen to be looking at, but as a constellation of light and shadow. While meditating, space is not carved up things that occupy reality. Rather you’re paying attention to raw sensation, raw visual data. Concepts like ‘chair’, ‘table’, ‘window’, and such fall away.

If concepts like chair and table fall away, it’s easy to see that selves fall away as well. If there aren’t chairs, then there likely aren’t people. But how could one come to know that there are no chairs through meditation?

Meditation does show that our notion of things depends on having concepts. But this doesn’t show that there are no such things. An infant doesn’t have concepts like a table, but this is because it’s stupid not because its closer to reality.

A more suggestive way that meditation can come to show that you have no self, is that we learn that our sense of being a self is merely a sensation. This self feels like something that has thoughts. But, through training attention, we notice that our sense of self can’t have thoughts. It is in the mental field, it cannot contain it. But this argument confuses the sense of self, with the self. These are distinct. Suppose that what I am is a human organism. Animals can have thoughts and host a mental field! This argument makes a similar mistake to indirect realists, confusing how we experience something with the thing itself.

So much for two arguments for the ontological relevance of meditation.

Ultimately, we cannot learn about the world from meditation because of cognitive penetration. The idea is that our experience is shaped by our beliefs and desires. The philosopher Evan Thompson describes going to a retreat, in Why I Am Not A Buddhist:

Nevertheless, I couldn’t help thinking throughout the retreat that what was happening didn’t match the rhetoric of “learning to see things as they are.” We were given a system of concepts to apply to our experience as we practiced meditation. Some of the concepts were seemingly everyday ones like “sensation,” “feeling,” “attention,” and “intention,” but they were tied to Buddhist concepts like “moment-to-moment arising,” “impermanence,” “mindfulness,” “not-self,” and “karma.”

Though the meditation practice was ostensibly nonconceptual, it happened in an ecosystem that was already populated by concepts. This shaped the experience of the “nonconceptual” practice.

Consider the variety of things people say you can come to learn through meditation. Though the traditional Buddhist view is the no-self one, there are others who say that you may come to extend the self through meditation. You could come to learn that everything is in the mind, but you could also come to learn that there is no mind and everything is out in the world. What you learn is contingent on how meditation is framed and the environment one meditates in.

In this way, meditative experiences are similar to religious experiences. What the experience is depends on how it is conceptualized. A Christian experiences the love of God, a Buddhist Nirvana, an atheist nature.

This isn’t to deny that meditation is valuable. However, you should invest in how you and the people around frame it. What meditation is useful for, is for training particular skills and generating experiential, perspectival knowledge. It’s one thing to know something abstractly, another thing to experience it.

But it’s not ontologically relevant.

You See The Territory Or Nothing At All

This post is about the philosophy of perception, what makes a philosophical debate substantive, and cognitive qualia.

Consider two views about the nature of perception and the world: on indirect realism we perceive a veil, we only indirectly get at the world through this veil. On direct realism, we perceive the world. The veil is lifted and we are in direct contact with the world.

On indirect realism what we perceive are mental objects like sense data. This is the veil that stands between us and the world. We must infer the existence of the world.

You can visualize the two views in the following ways:

indirect: perceiver => sense data => world
direct: perceiver => world

indirect: perceiver => map => territory
direct: perceiver => territory

indirect: perceiver => veil => world
direct: perceiver => world

Indirect realism gets its intuitive appeal from the idea of hallucination. If I have the experience of a red elephant in my room then I am perceiving my visual field, not a red elephant in the room. The object of my perception is the appearance of a red elephant, nothing more. We can visualize this as:

perceiver => experience of a red elephant => 

The last arrow points to nothing because there is no red elephant in the world. Here’s how the direct realist would visualize this case:

perceiver => 

The direct realist says that you don’t perceive anything when you hallucinate. You either see the territory or nothing at all.

Despite the intuitive appeal of indirect perception, the direct view is better. Using our visualization, you can get at this by stepping back and considering the question that a theory of perception should answer: what is the object of one’s perception? In the case where perception is accurate, the answer is the world. But for the indirect realist, the object of perception is a mental object, like sense data.

But how is the object of perception perceived? What is it perceived through?

On direct realism the object of perception is perceived through mental states, which we can call experience. Returning to the visualization, the `=>` is experience.

perceiver => world = perceiver experiences the world

The indirect realist cannot say this. This is because the object of perception is a mental object. Which is to say, it is experience. On this, picture one only experiences experiences. You can visualize this as follows:

indirect: perceiver => experience => world = perceiver experiences experience

Of course, there’s a sense in which you can experience experiences. I can experience an AirBnB experience. But when we say that, experience means two different things. What the indirect realist is committed to is the idea that one perceives mental states through other mental states at the same moment. This isn’t right, there’s just experience. Not experiences of experience. The mistake that the indirect realist makes is that they confuse the object of perception with the vehicle of perception, what was is aware of with how one is aware. We’re aware through experience, not just aware of things in our head.

You might be wondering what is going on in this debate. Is it substantive? Sometimes we debate things that are substantive, like whether there is a God or not. On other occasions, it seems like we are playing verbal or conceptual games.

I think this is a case where we’re playing conceptual games.

Here are two tests for whether something is a conceptual debate or not.

First the taboo test. Try to carry on the debate in other terms. If you can’t do this, there’s a good chance you’re arguing over the meaning of words or a concept. Here’s how this would look like in the God debate:

Before taboo
atheist: there is no God
theist: there is a God

After
atheist: there is no supernatural triomni person
theist: there is a supernatural triomni person

There's clearly a substantive debate here. Whether or not there is a supernatural triomni person is not a linguistic matter.

How does it go in the perception debate?

Before taboo
direct: the object of perception is the world
realist: the object of perceptions are mental states 

After
direct: mental states are not the object of awareness
indirect: mental states are the the object of awareness

This, of course, pushes the debate from perception to awareness. This may game will be played a few rounds, but it’s not clear to me how to proceed once the words like perception, awareness, contents of experience, and acquaintance are tabooed. Which is one would expect if one is only arguing over the purview of a concept. Both parties agree that we have experiences that are candidates for perception. The direct realist doesn’t deny that we can have the experience of a red elephant suddenly appearing in our rooms. What they do deny is that there would be any perception going on in that case. But it's difficult to restate this debate without using perception or its synonyms.

A shorter way to determine whether a debate is purely conceptual is to ask whether it differentiates between different ways for the concrete world to be. Consider debates over the following propositions:

1. Pluto is a planet
2. If something is a planet, then it clears its planetary orbit

If two parties are disagreeing about whether Pluto is a planet, they are likely disagreeing over whether Pluto meets the requirements for being a planet, like whether it clears the neighborhood around its orbit. The concrete world would be one way if Pluto cleared its neighborhood, it would be another way if it didn’t.

If two parties disagreed over the second proposition, then they’d be debating the requirements for being a planet. This is a conceptual matter. The concrete world would not change if the proposition were true or not.

Now, move to the debate over the nature of perception. We have two propositions:

3. Charles perceives a red elephant
4. If someone perceives X, then X is a mental state

Does the first differentiate different ways for the concrete world to be? It depends. If what perception means is having a mental state that includes the appearance of a red elephant, then yes, it does. If the proposition were true, then Charles has a mental state that includes a red elephant. If it were false, then he has a different mental state. What it would be like to be Charles would be different in the two cases.

True: Charles has a mental state that includes the appearance of a red elephant.
False: Charles has a different mental state

Likewise, if what perception means is to accurately represent the world, then this to differentiates between different ways for the concrete world to be. In one case, there is a red elephant and in another case there isn’t. The concrete world is different in both cases:

True: there’s a red elephant
False: there’s no red elephant

The debate over the nature of perception isn’t about whether there are red elephants or not. Nor is it about whether one can have the mental state that would be characterized by the appearance of a red elephant.

Instead, it’s about whether perception has as its object the world or some mental state. Which is to say it’s about propositions like this:

If someone perceives X, then X is a mental state

Whether this proposition is true or not will make no difference to the concrete world. Which indicates that this debate is merely conceptual.

Cognitive Qualia

I used to think that debates over concepts didn’t matter. Of course, conceptual clarity is useful because it eases communication and carving up the world neatly makes it easier to handle.

But the debates don’t matter in the sense that you aren’t figuring out anything fundamental about the concrete world. Whether or not there are quarks, whether ants are sentient, or whether egalitarian societies are more productive -- these debates matter. They get at different ways for the world to be in a way that purely conceptual debates do not.

But there’s another way in which these debates matter, something that took me a few years to realize.

I was reminded of this by hearing Eric Weinstein say the following to Josh Wolfe:

The thing that I am talking to now is the projection of you inside of my head. That comes from the stimulation. I mean, I believe that I have eyes, I don't want to get completely Jiggy but just assume that reality is the standard picture of reality. I see you as being across the room for me, but that's not the thing that it's really happening, what’s really happening, if everything I know to be is correct, Is that the thing across the room for me is generated a model in my mind, which is the only thing I've ever interacted with.

This is wrong. Eric spoke to Josh Wolfe, not a model of Josh Wolfe.

But that’s not the main point here, the main idea is that what it’s like to be a direct realist can differ radically from what it’s like to be an indirect realist. In other words, there’s such a thing as cognitive qualia. Qualia refers to the subjective character of mental states. There’s something it’s like to be sad, overjoyed, feel pain, feel pleasure. Likewise, there’s something it’s like to believe that you can see the territory, something else to believe that you’re stuck to seeing the map.

Robin Hanson has called out how beliefs are basically just like clothes. They have a social and functional role:

Functionally, beliefs inform us when we choose our actions, given our preferences. But many of our beliefs are also social, in that others see and react to our beliefs. So beliefs can also allow us to identify with groups, to demonstrate our independence and creativity, and to signal our wealth, profession, and social status.


This picture is insightful, but incomplete. Beliefs have an experiential role as well.

On Being Angry Forever

Why do we concern ourselves with conflict and plotting? That man you are angry with – can you wish for him anything worse than death? He is going to die without your doing a thing.

Seneca, On Anger

Agnes Callard has an argument for why you can be angry forever. I think it’s wrong, but it useful for thinking about the nature of emotions:

1. There are reasons to be angry
2. Reasons to be angry do not go away.
3. If reasons to be angry do not go away, then all things considered, you have reasons to remain angry forever.
4. Therefore, all things considered, you have reasons to be angry forever.

Callard calls this the argument from grudges:

Once you have a reason to be angry, you have a reason to be angry forever.

Many people accept (1). (2) is more surprising. Here’s Callard on the intuition:

The tendency to cling to anger through apologies and recompense, for years sometimes and to the detriment of all parties concerned, is routinely dismissed as irrational...

But this idea ignores the fact that there are reasons to remain angry. And the reasons are not hard to find: they are the same reasons as the reasons to get angry in the first place. Apologies, restitution, and all the rest do nothing to cancel or alter the fact that I stole, nor the fact that I ought not to have stolen.

In other words, reasons to be angry do not go away. If you’re angry at the fact that someone stole from you, the act that they stole does not go away.

Grudges are not that different from other emotions, like gratitude and grief. If you’re grateful that someone helped you in a time of need, then the fact that they helped you in a time of need is something you can be grateful about forever. If you experience grief at the death of a loved one, then you have a reason to grieve for all time.

This comparison highlights an asymmetry between positive and negative emotions. It’s reasonable to be grateful forever, but not to hold grudges or grieve forever.

This asymmetry seems fundamentally right to me.

Emotions are instrumentally useful. They shape our experiences, actions, and model of the world. When we have a reason to be angry, it’s not just because someone stole from us. It’s because being angry will help us stand up for our rights. In my view, by responding in a  particular way, your reasons are dominated by your past, what you’d like to do and who you want to be. Callard’s view assumes that the reasons to have some emotional response are merely determined by the facts we are reacting to. The problem with this view is that it's not clear what would determine what makes some emotions fit to a situation other than facts about us, our needs and past, and the situation.

This explains why anger fades overtime. Emotion fades as it becomes useful. There is no point at being angry at someone when doing so will make no difference. It also explains the asymmetry between positive and negative emotions. Prompting positive emotions by remembering reasons to be grateful or happy is often useful for crafting who we want to be and how we want to act.

Anger is only useful when it’s useful, and for most of us, that’s increasingly rare.

Endurable Startup Opportunities From Covid19

Covid19 has rapidly and seriously changed the American economy and culture. How endurable are these changes? If these changes are endurable, there will be many Covid19 related post-lockdown opportunities for startups. If not, perhaps we will reset the world in a year or so and startup opportunities will be the exact same as they were in 2019.

My sense is that we’ll see the acceleration of existing trends and a few serious opportunities, but that this can be overstated. If you’re looking for new ideas, don’t update too much on Covid19.

The lockdown has shaped many spheres of our life:

  • Social
    • More games with friends and family over the internet.
    • More group chats.
    • We interact with and may trust strangers a lot less.
    • Demand for better group events. What happened to concerts?
  • Entertainment
    • We have more time and we can’t spend as much of it outside. More Netflix and video games.
  • Wellness
    • We have more time and don’t want to spend all of it on entertainment. Exercise apps like Strava, Nike Run and meditation apps like Calm and Stoa are in.
    • There’s a lot of uncertainty and discomfort, I was told that psychics saw an exceptional boost when relief checks were delivered. I believe it.
  • Work
    • More remote work.
    • Demand for tools and processes that support remote work.
  • E-commerce
    • Medical supplies, food and beverages are in. Fashion and luxury goods are out.
    • Subscription and convenience services are up.
  • Supply chains
    • We’ve learned that many of our supply chains were fragile. Keeping supply chains in America is likely more expensive for both suppliers and consumers, but we’d be more robust as a nation.
  • Science
    • We’ve increased the status for work on pandemic preparedness.
  • State Augmentation
    • Startups and big companies are supplying the state with more masks and tracking programs. The sort of thing that authoritarian governments would have the state do, America has private companies do.

In the short term, there are many useful things to build and contribute to.

In the longer term, I expect that many of these spheres will return to their prior shape. However, there’s promising work to be done and products to built.

  • Social
    • In person interaction is still much better than chatting over Hangouts. We largely want to date, dance, and talk in person. Things don’t have to be this way, but I haven’t seen solutions to this yet.
  • Entertainment
    • Already dying industries like movie theaters have had their death accelerated.
  • Wellness
    • I expect that the small spike is just that and doesn’t indicate a trend. Remote wellness services are not that sticky. The dark secret behind many successful wellness services is that they’re making money off of people who are subscribed but don’t use their product. Nike Run cannot replace the social and motivational goods of CrossFit yet. There’s no evidence to believe that this will change.
  • Work
    • More and more tech companies and startups have already been experimenting with remote work. I expect that trend will continue, but I don’t know whether Covid19 will speed it up or slow it down. On one hand, companies are likely learning that they can continue to be just as productive while remote. On the other, this learning is correlated with unpleasant loneliness and isolation. This association is real and may not go away.
  • E-commerce
    • I’d bet on subscription and convenience services. I’m long Amazon and the ecosystem it requires.
  • Supply chains
    • I wouldn’t be surprised to see more protectionism and a push for bringing supply chains back to America. But a major one? Unlikely.
  • Science
    • There’s a lot of exciting and important work to do here. More and more people will be thinking about pandemic preparedness which is great.
  • State Augmentation
    • The exact shape of this is unclear, but this area is really promising for companies.

I expect that we’ll see an unbundling of services that are stretched across these domains. Right now Zoom is used for work, socializing, and entertainment. Services that specialize in socializing and services that specialize with different kinds of work is promising. One way this can go is that existing services bring video conferencing into their software. There’s space for entirely new products here. Facebook Messenger Rooms is a stab at this.

My sense of how things may go post-lockdown is not set in stone. I expect that I got at least one of the above wrong, someone will make something I won’t expect.

If there’s a cultural change that I hope persists, it’s that America raise the status of building and surviving.