- Consider a plethora of possibilities--dragon eyes.
---- What would have to be true for __ to happen?
---- What would have to be true for __ NOT to happen?
- Look for additional perspectives to synthesize with their own
- Analyze their assumptions critically.
---- Would I be convinced by this I was someone else?
---- Where are the holes?
---- Assume they are wrong
- Are okay with uncertainty, not limited to binary thinking
- Update their forecast constantly
- Anchor on an outside view, a base rate (how common is X in the big picture)
- Embraces probabilistic thinking
- Growth mindset
- Practices
- Learns and adjusts, okay getting things wrong, desires 'light' to expose mistakes so they grow
- Stable
- Cultivates relationships where they can safely be challenged, challenge, be wrong, get help and help others.
- Giver
Q2:
- Less updating of forecasts
- Less granular in predictions
- Fixed mindset
- Prone to easier route, less critical thinking and avoidance of the harder question
- One perspective analysis, stuck on a Big Idea
- Prefers certainty of uncertainty
- Doesn't self-critique
- Anchors on the specific story and looks for facts to justify it (inside view first)
- Under/overreacts to new information, big swings
- More ego in process
- Doesn't seek feedback on accuracy of forecast
Q3:
What information would allow me to answer this question?
In 2030, how many cars will be sold?
How many car buyers will want EV?
How many will choose a Tesla?
Develop baselines:
How many cars are sold last year? ---- ~90 million
How many of those where EV's last year? ---- ~ 20 million
How many were Tesla? ---- ~1.5 million
Baseline: 22% of cars are sold globally are EV. Tesla's current market share is of EVs is 7.5%, but 1.6% of world EV. Currently, Tesla doesn’t dominate the global EV market.
2030: Would the baseline numbers change? How/Why?
Will the number of cars sold per year change? ---- Cars are big ticket items. I don't see a massive increase. If anything, economic challenges could pressure these numbers down. Tesla demand: +/-1% Plausible
What's the EV car trend been like over the past 5-10 years? Growing? Decreasing? Leveling out? ---- Loss of tax credits in the US decreased demand for EV and there are still looming challenges for EVs. However, I see more EV cars on the road and more infrastructure to support them, so I think it's growing. I don't think people will stop buying them altogether, but the pace of adoption may be slow for a while still. I don't think by 2030 there will yet be a sweeping increase. Tesla demand: + 0-0.5% Probable
How many EV buyers will choose a Tesla? ---- Tesla doesn't have global 'moat' on EV cars anymore. Other companies can produce as or more desirable EV cars (cheaper, comfortable, roomier, better batteries). Although Tesla does dominate, I've seen more variety in my area lately. My experience is US based where I see a lot of Tesla's out there in comparison to other companies. But I'm aware that globally there are other dominate brands that have gained market share. I don't see them getting a bigger share of the market. Tesla demand: +/- 0-0.5% Probable
What would have to be true for MORE than base? ---- US tax credits would be helpful. China consumer switching back to Tesla (not likely due to price point differences) Tesla demand: +3% Very unlikely
What would have to be true for LESS than base? ---- EV Cars from other countries become dominate in the US market. (plausible, but that's a steep slope) Tesla demand: -1% Less likely
Forecast: Telsa in 2030 has a market share of 0.5%-2%
Q4:
Many items are applicable. I'd like to develop deeper critical thinking in my thesis building process. Specifically, I'd like to work on developing more 'other hands' and negative thesis. I'd like to be more creative and dig to find the real questions, not just the surface bait and switch question. I want to add the outside view first so I'm anchoring closer to reality than just the 'story'.
Very important to try to understand where you can be wrong, as well as decide in advance which signs/new evidence should make you believe it's then more likely that you are wrong. This is where AI can be helpful as a 'devil's advocate' or a de-biaser.
Q1. The ability to think. Using appropriate tools and questions, having the ability to change their mind, have the ability to be wrong, looks at the situation from many perspectives - logic and psycho-logic, looks at situation from the view of the opponent, decision making is on a spectrum - the fox and hedgehog analogy, start with an outside view - a wider spectrum, have active open mindedness, use probabilistic thinking, unpacks questions into component parts, Fermi - separates knowable from unknowable, their forecasts are a product of careful thought and nuanced judgement. Uses more System 2 thinking. Updates forecasts more frequently and in smaller increments.
Q2, Doesn't consider alternative views, or a wide enough spectrum of viewpoint, and not comfortable being wrong, uses percentages as absolutes. Uses more system 1 thinking which introduces unchallenged biases. Basically thinking in the opposite of the qualities mentioned in question 1.
Q3. 2%. A lot of the world population doesn't have cars, so I would look at the percentages of the world population by location, and income levels. How many people already have cars and how many don't. How many of those that don't will be purchasing them in the next 5 years. How cars right now are electric and how may gas powered. How many people live lifestyles which would be appropriate to using electric cars. How much of the population have access to charge the vehicle. How many need to be able to travel long distances which would be more suited for gas powered vehicles.
Q4. Improve both the quality and scope of the questions I ask. Being more conscious of the biases I have and moving more into System 2 style thinking. Use Charlie's mental models - invert, always invert.
Question 1: What are the common characteristics of superforecasters?
Need for Cognition: Without this, no one could possibly be a good forecaster; it's the driver/motivation for putting in the work, because for those with an NFC, it's not work, it's fun.
(I did the NFC test of 18 questions; it took me about 30 seconds to get a 100% score – I have a serious NFC bias in pretty much everything I do, so I know what a strong driver it can be...)
But NFC is not enough to be a good superforecaster; there are other qualities, some of which are hard for those who are used to being right and being the smartest person in the room:
Analytical, able to quantify in depth if necessary
Curious, able to think laterally and come at problems from multiple angles – dragonfly eyes
Cautious: nothing is certain.
Flexible: able to change their mind without losing sight of their original analysis: a Baesian approach
Self-critical: always prepared to re-examine assumptions and challenge thesis with antithesis to achieve synthesis.
Able to break down complex problems into simpler steps: Fermi approach to the number of piano tuners in Chicago
Able to examine each step and relentlessly improve, if necessary, many times
Able to conduct postmortems, admit to mistakes and learn from them,
Able to work in teams to get better outcomes
Grit: the power to get it catastrophically wrong multiple times and still keep trying.
Question 2: What are the common characteristics of the average/not very good forecasters?
Arrogance
Certainty
Binary thinking: 100% chance of happening, or 100% chance of not happening.
Vague parameters: for example, no timelines for a prediction
No feedback or holding to account on predictions
Allowing beliefs such as predestination to affect worldview.
Having an intellectual framework for viewing the world that is set: free-marketeers, hawks, doves, bulls, bears
Question 3: By yourself, predict Tesla's market share of all cars globally by 2030. Explain your reasoning.
starting with global car sales. The number of units sold and forecast in millions is:
2015 72.7
2016 76.5
2017 78.4
2018 77.7
2019 73.8
2020 62.7
2021 65.2
2022 65.1
2023 72.8
2024 74.6
2025 83.0
2026 84.0
2027 85.0
2028 87.0
2029 89.0
2030 91.0
This is a reasonably slow increase if you ignore the dip during covid, so it seems a good estimate. More than a 2% deviation from this would be a long way from the line and a marked change in trend.
Tesla sold 1.6m units in 2025. Its current market share of all vehicles sold is therefore 1.9%.
Around 20% of cars globally are EVs, so about a 10% market share of EVs.
This is Tesla's estimate as a consensus from its website of around 20 broker forecasts for the next 4 years.
2026: 1,750,243
2027: 2,010,459
2028: 2,350,451
2029: 3,019,902
2030: 3,900,000 (UBS estimate, relatively conservative). Some are much higher.
So our anchor forecast for market share by unit is 4.2% from published data.
Do we believe these projections? Do they need modifying?
These are the historical volumes produced by Tesla.
2018 245,240
2019 367,500
2020 499,550
2021 936,000
2022: 1,313,851
2023: 1,808,581
2024: 1,789,226
2025: 1,636,129
2025 sales by region
Model US EEA/Europe* China
Model Y: 357,528, 149,805, 425,337
Model 3: 192,440, 85,393, 200,361
The unit volume has been falling for the last 2 years, and the competition is rising fast. BYD in particular is moving extremely rapidly, with its latest models competing at both the bottom end with sub-$10K models in China and 1,000 km range high-end models. Tesla sold 625K units in China in 2025, more than a third of total sales. I think it unlikely that this will continue to grow rapidly, and European sales have also been very weak due to poor publicity.
For Tesla to hit these numbers, it will need to turn this round. That means significant growth in new areas like robotaxis and pushing new models successfully. All this while the CEO is concentrating on an enormous IPO of SpaceX this year.
Assume it holds market share in China, regains momentum in Europe and makes modest growth in the US.
US 800,000
Europe 450,000
China 625,000
Giving a total of 1,875,000 and a market share of 2% of units shipped in 2030.
So my answer is a range between 2% and 4.2% of units sold globally in 2030. Where it is in that range will depend on whether it cracks robotaxis in a significant way and whether Elon Musk spends enough time on it. My bet is on the lower end of the forecast.
Question 4: What ideas from this book can you incorporate into your own investing?
One of the most difficult aspects of DCF models is that they illustrate just how much of a company's value lies far into the future. So while we may feel uncomfortable making long-range predictions, when we buy a company, we are doing just that. In a 10-year forecast, typically around 50% of the value lies in the terminal value of any good-quality company, so by buying it at that price, we are assuming it will be around and valuable in 10 years. Working out the forecast more than 3 years out is where we can either see value where others cannot or vice versa. This kind of thinking is what is needed to build those longer-term forecasts and is where our view of value can look further ahead than the markets. It also should make us pause and think both hard and humbly about whether we can actually say anything meaningful about company forecasts so far into the future. For some companies this feels more certain than for others, and this also we should take into account in what we choose to buy and at what price.
What kept coming to my mind is the following well known problem.
Suppose you're on a game show, and you're given the choice of three doors: Behind one door is a car; behind the others, goats. You pick a door, say No. 1, and the host, who knows what's behind the doors, opens another door, say No. 3, which has a goat. He then says to you, "Do you want to pick door No. 2?" Is it to your advantage to switch your choice?
As usual, I use your questions to think more about situations, so forgive me if much of the below is obvious to you; answering the question in this way is my way of thinking the problem through and trying to learn.
First off, you are right; anchoring is dangerous, and I should have worked out the answer without reference to broker forecasts. But the reason I did so without thinking enough is that I genuinely think that the opinions of others need to be taken into consideration. This niggled at me, so here is the thought chain as to why it did.
Most problems/games can be divided into two classes: one where there is perfect information, where everyone's knowledge is equal, and all the rest, where not all the information is visible to all parties. Examples of the first are tic-tac-toe, checkers, chess, go and the like. Examples of the second are poker and most real-life situations like the stock market.
Humans find it extremely difficult to evaluate problems and knowledge from all points of view and "put themselves in someone else's shoes" to assess what they might know. It's a skill that most people don't even begin to have until they are teenagers, and some of the hardest problems with perfect information that you can devise involve this concept. It's particularly difficult if there is a probabilistic estimate attached to different possible types of other humans, like the famous Monty Hall problem I quoted.
In the problem you set us we have to estimate something that is intrinsically hard to know and with many factors affecting it; that is the point of the problem. There are two approaches to it.
1) Estimate using visible knowledge from what you find out and your own existing knowledge and experience. This is the instinctive and intuitively right way to do it, as we are in a perfect information world – we know what we know, and we don't know what we don't know. This is what you asked us to do and I should have done.
2) Estimate taking into account the opinions of others. Here it gets much harder. We now have to estimate what we don't know, and we have to estimate whether others might be in a position to know more than us. We also have to estimate whether others are saying what they really think or what they are incentivised to say, which may or may not be the same thing. We should also consider whether they might be influencing each other with groupthink, of which there are plenty of famous examples, like the Bay of Pigs or WMDs. This feels much more like playing poker.
Producing an opinion without taking any of this into account feels dangerous to me. The automatic assumption is that we know better and that our view will be less "biased" or "anchored" if we ignore the opinions of others.
The ideal solution is to keep digging so that your information increases and the unknown part hopefully decreases, and your estimate gets better and more certain over time: the super forecaster approach. The trouble is that there are 3 categories of knowledge here: the things I know, the things I don't know, and the things I don't know I don't know. It's the third category that is the dangerous one, the 'unknown unknowns', as engineers call it.
In practice I am acutely aware that I am only one human out of 8 billion and that my knowledge of all the factors that go into Tesla's sales over the next few years is imperfect to say the least and that I'm not a super forecaster. Given this position, it seems logical to me to allow that, at least to start with, I should pay some attention to what others think and only discard it when I have enough evidence to satisfy me that they are wrong.
I read Superforecasters as it was released. One of the best reads in my 66 years. Now, as with most things, his human work is about to be subsumed by artificial intelligence. No doubt, AI is at his elbow.
Q1:
- Curious, likes to learn
- Question beliefs, actively open-minded
- Consider a plethora of possibilities--dragon eyes.
---- What would have to be true for __ to happen?
---- What would have to be true for __ NOT to happen?
- Look for additional perspectives to synthesize with their own
- Analyze their assumptions critically.
---- Would I be convinced by this I was someone else?
---- Where are the holes?
---- Assume they are wrong
- Are okay with uncertainty, not limited to binary thinking
- Update their forecast constantly
- Anchor on an outside view, a base rate (how common is X in the big picture)
- Embraces probabilistic thinking
- Growth mindset
- Practices
- Learns and adjusts, okay getting things wrong, desires 'light' to expose mistakes so they grow
- Stable
- Cultivates relationships where they can safely be challenged, challenge, be wrong, get help and help others.
- Giver
Q2:
- Less updating of forecasts
- Less granular in predictions
- Fixed mindset
- Prone to easier route, less critical thinking and avoidance of the harder question
- One perspective analysis, stuck on a Big Idea
- Prefers certainty of uncertainty
- Doesn't self-critique
- Anchors on the specific story and looks for facts to justify it (inside view first)
- Under/overreacts to new information, big swings
- More ego in process
- Doesn't seek feedback on accuracy of forecast
Q3:
What information would allow me to answer this question?
In 2030, how many cars will be sold?
How many car buyers will want EV?
How many will choose a Tesla?
Develop baselines:
How many cars are sold last year? ---- ~90 million
How many of those where EV's last year? ---- ~ 20 million
How many were Tesla? ---- ~1.5 million
Baseline: 22% of cars are sold globally are EV. Tesla's current market share is of EVs is 7.5%, but 1.6% of world EV. Currently, Tesla doesn’t dominate the global EV market.
2030: Would the baseline numbers change? How/Why?
Will the number of cars sold per year change? ---- Cars are big ticket items. I don't see a massive increase. If anything, economic challenges could pressure these numbers down. Tesla demand: +/-1% Plausible
What's the EV car trend been like over the past 5-10 years? Growing? Decreasing? Leveling out? ---- Loss of tax credits in the US decreased demand for EV and there are still looming challenges for EVs. However, I see more EV cars on the road and more infrastructure to support them, so I think it's growing. I don't think people will stop buying them altogether, but the pace of adoption may be slow for a while still. I don't think by 2030 there will yet be a sweeping increase. Tesla demand: + 0-0.5% Probable
How many EV buyers will choose a Tesla? ---- Tesla doesn't have global 'moat' on EV cars anymore. Other companies can produce as or more desirable EV cars (cheaper, comfortable, roomier, better batteries). Although Tesla does dominate, I've seen more variety in my area lately. My experience is US based where I see a lot of Tesla's out there in comparison to other companies. But I'm aware that globally there are other dominate brands that have gained market share. I don't see them getting a bigger share of the market. Tesla demand: +/- 0-0.5% Probable
What would have to be true for MORE than base? ---- US tax credits would be helpful. China consumer switching back to Tesla (not likely due to price point differences) Tesla demand: +3% Very unlikely
What would have to be true for LESS than base? ---- EV Cars from other countries become dominate in the US market. (plausible, but that's a steep slope) Tesla demand: -1% Less likely
Forecast: Telsa in 2030 has a market share of 0.5%-2%
Q4:
Many items are applicable. I'd like to develop deeper critical thinking in my thesis building process. Specifically, I'd like to work on developing more 'other hands' and negative thesis. I'd like to be more creative and dig to find the real questions, not just the surface bait and switch question. I want to add the outside view first so I'm anchoring closer to reality than just the 'story'.
Very important to try to understand where you can be wrong, as well as decide in advance which signs/new evidence should make you believe it's then more likely that you are wrong. This is where AI can be helpful as a 'devil's advocate' or a de-biaser.
Q1. The ability to think. Using appropriate tools and questions, having the ability to change their mind, have the ability to be wrong, looks at the situation from many perspectives - logic and psycho-logic, looks at situation from the view of the opponent, decision making is on a spectrum - the fox and hedgehog analogy, start with an outside view - a wider spectrum, have active open mindedness, use probabilistic thinking, unpacks questions into component parts, Fermi - separates knowable from unknowable, their forecasts are a product of careful thought and nuanced judgement. Uses more System 2 thinking. Updates forecasts more frequently and in smaller increments.
Q2, Doesn't consider alternative views, or a wide enough spectrum of viewpoint, and not comfortable being wrong, uses percentages as absolutes. Uses more system 1 thinking which introduces unchallenged biases. Basically thinking in the opposite of the qualities mentioned in question 1.
Q3. 2%. A lot of the world population doesn't have cars, so I would look at the percentages of the world population by location, and income levels. How many people already have cars and how many don't. How many of those that don't will be purchasing them in the next 5 years. How cars right now are electric and how may gas powered. How many people live lifestyles which would be appropriate to using electric cars. How much of the population have access to charge the vehicle. How many need to be able to travel long distances which would be more suited for gas powered vehicles.
Q4. Improve both the quality and scope of the questions I ask. Being more conscious of the biases I have and moving more into System 2 style thinking. Use Charlie's mental models - invert, always invert.
Question 1: What are the common characteristics of superforecasters?
Need for Cognition: Without this, no one could possibly be a good forecaster; it's the driver/motivation for putting in the work, because for those with an NFC, it's not work, it's fun.
(I did the NFC test of 18 questions; it took me about 30 seconds to get a 100% score – I have a serious NFC bias in pretty much everything I do, so I know what a strong driver it can be...)
But NFC is not enough to be a good superforecaster; there are other qualities, some of which are hard for those who are used to being right and being the smartest person in the room:
Analytical, able to quantify in depth if necessary
Curious, able to think laterally and come at problems from multiple angles – dragonfly eyes
Cautious: nothing is certain.
Flexible: able to change their mind without losing sight of their original analysis: a Baesian approach
Self-critical: always prepared to re-examine assumptions and challenge thesis with antithesis to achieve synthesis.
Able to break down complex problems into simpler steps: Fermi approach to the number of piano tuners in Chicago
Able to examine each step and relentlessly improve, if necessary, many times
Able to conduct postmortems, admit to mistakes and learn from them,
Able to work in teams to get better outcomes
Grit: the power to get it catastrophically wrong multiple times and still keep trying.
Question 2: What are the common characteristics of the average/not very good forecasters?
Arrogance
Certainty
Binary thinking: 100% chance of happening, or 100% chance of not happening.
Vague parameters: for example, no timelines for a prediction
No feedback or holding to account on predictions
Allowing beliefs such as predestination to affect worldview.
Having an intellectual framework for viewing the world that is set: free-marketeers, hawks, doves, bulls, bears
Question 3: By yourself, predict Tesla's market share of all cars globally by 2030. Explain your reasoning.
starting with global car sales. The number of units sold and forecast in millions is:
2015 72.7
2016 76.5
2017 78.4
2018 77.7
2019 73.8
2020 62.7
2021 65.2
2022 65.1
2023 72.8
2024 74.6
2025 83.0
2026 84.0
2027 85.0
2028 87.0
2029 89.0
2030 91.0
This is a reasonably slow increase if you ignore the dip during covid, so it seems a good estimate. More than a 2% deviation from this would be a long way from the line and a marked change in trend.
Tesla sold 1.6m units in 2025. Its current market share of all vehicles sold is therefore 1.9%.
Around 20% of cars globally are EVs, so about a 10% market share of EVs.
This is Tesla's estimate as a consensus from its website of around 20 broker forecasts for the next 4 years.
2026: 1,750,243
2027: 2,010,459
2028: 2,350,451
2029: 3,019,902
2030: 3,900,000 (UBS estimate, relatively conservative). Some are much higher.
So our anchor forecast for market share by unit is 4.2% from published data.
Do we believe these projections? Do they need modifying?
These are the historical volumes produced by Tesla.
2018 245,240
2019 367,500
2020 499,550
2021 936,000
2022: 1,313,851
2023: 1,808,581
2024: 1,789,226
2025: 1,636,129
2025 sales by region
Model US EEA/Europe* China
Model Y: 357,528, 149,805, 425,337
Model 3: 192,440, 85,393, 200,361
The unit volume has been falling for the last 2 years, and the competition is rising fast. BYD in particular is moving extremely rapidly, with its latest models competing at both the bottom end with sub-$10K models in China and 1,000 km range high-end models. Tesla sold 625K units in China in 2025, more than a third of total sales. I think it unlikely that this will continue to grow rapidly, and European sales have also been very weak due to poor publicity.
For Tesla to hit these numbers, it will need to turn this round. That means significant growth in new areas like robotaxis and pushing new models successfully. All this while the CEO is concentrating on an enormous IPO of SpaceX this year.
Assume it holds market share in China, regains momentum in Europe and makes modest growth in the US.
US 800,000
Europe 450,000
China 625,000
Giving a total of 1,875,000 and a market share of 2% of units shipped in 2030.
So my answer is a range between 2% and 4.2% of units sold globally in 2030. Where it is in that range will depend on whether it cracks robotaxis in a significant way and whether Elon Musk spends enough time on it. My bet is on the lower end of the forecast.
Question 4: What ideas from this book can you incorporate into your own investing?
One of the most difficult aspects of DCF models is that they illustrate just how much of a company's value lies far into the future. So while we may feel uncomfortable making long-range predictions, when we buy a company, we are doing just that. In a 10-year forecast, typically around 50% of the value lies in the terminal value of any good-quality company, so by buying it at that price, we are assuming it will be around and valuable in 10 years. Working out the forecast more than 3 years out is where we can either see value where others cannot or vice versa. This kind of thinking is what is needed to build those longer-term forecasts and is where our view of value can look further ahead than the markets. It also should make us pause and think both hard and humbly about whether we can actually say anything meaningful about company forecasts so far into the future. For some companies this feels more certain than for others, and this also we should take into account in what we choose to buy and at what price.
One danger in looking at consensus numbers prior to forming your own forecast is a strong anchoring effect, whether you realize it or not.
As always your comments spark some thoughts.
What kept coming to my mind is the following well known problem.
Suppose you're on a game show, and you're given the choice of three doors: Behind one door is a car; behind the others, goats. You pick a door, say No. 1, and the host, who knows what's behind the doors, opens another door, say No. 3, which has a goat. He then says to you, "Do you want to pick door No. 2?" Is it to your advantage to switch your choice?
My initial take is that these are different situations
As usual, I use your questions to think more about situations, so forgive me if much of the below is obvious to you; answering the question in this way is my way of thinking the problem through and trying to learn.
First off, you are right; anchoring is dangerous, and I should have worked out the answer without reference to broker forecasts. But the reason I did so without thinking enough is that I genuinely think that the opinions of others need to be taken into consideration. This niggled at me, so here is the thought chain as to why it did.
Most problems/games can be divided into two classes: one where there is perfect information, where everyone's knowledge is equal, and all the rest, where not all the information is visible to all parties. Examples of the first are tic-tac-toe, checkers, chess, go and the like. Examples of the second are poker and most real-life situations like the stock market.
Humans find it extremely difficult to evaluate problems and knowledge from all points of view and "put themselves in someone else's shoes" to assess what they might know. It's a skill that most people don't even begin to have until they are teenagers, and some of the hardest problems with perfect information that you can devise involve this concept. It's particularly difficult if there is a probabilistic estimate attached to different possible types of other humans, like the famous Monty Hall problem I quoted.
In the problem you set us we have to estimate something that is intrinsically hard to know and with many factors affecting it; that is the point of the problem. There are two approaches to it.
1) Estimate using visible knowledge from what you find out and your own existing knowledge and experience. This is the instinctive and intuitively right way to do it, as we are in a perfect information world – we know what we know, and we don't know what we don't know. This is what you asked us to do and I should have done.
2) Estimate taking into account the opinions of others. Here it gets much harder. We now have to estimate what we don't know, and we have to estimate whether others might be in a position to know more than us. We also have to estimate whether others are saying what they really think or what they are incentivised to say, which may or may not be the same thing. We should also consider whether they might be influencing each other with groupthink, of which there are plenty of famous examples, like the Bay of Pigs or WMDs. This feels much more like playing poker.
Producing an opinion without taking any of this into account feels dangerous to me. The automatic assumption is that we know better and that our view will be less "biased" or "anchored" if we ignore the opinions of others.
The ideal solution is to keep digging so that your information increases and the unknown part hopefully decreases, and your estimate gets better and more certain over time: the super forecaster approach. The trouble is that there are 3 categories of knowledge here: the things I know, the things I don't know, and the things I don't know I don't know. It's the third category that is the dangerous one, the 'unknown unknowns', as engineers call it.
In practice I am acutely aware that I am only one human out of 8 billion and that my knowledge of all the factors that go into Tesla's sales over the next few years is imperfect to say the least and that I'm not a super forecaster. Given this position, it seems logical to me to allow that, at least to start with, I should pay some attention to what others think and only discard it when I have enough evidence to satisfy me that they are wrong.
I read Superforecasters as it was released. One of the best reads in my 66 years. Now, as with most things, his human work is about to be subsumed by artificial intelligence. No doubt, AI is at his elbow.