Upside risk
blog.samaltman.com
blog.samaltman.com
And because no one can predict the future, they must give the same onerous terms to that one "moonshot success" which will give them their 3000x return. They simply don't know which one it is in advance.
Agreed. That being said, putting all your liquid capital to work in one round is pretty imprudent, no?
Another big reason would for recruiting. I've seen that's become increasingly common that job offers include options as prices instead of percentages. Eg you'll get $500,000 in stock based on the last round valuation as opposed to saying 0.5%. This is increasingly common in later stage companies and it seems to be very effective. Now companies in early stage with high valuations can do the same.
Deleted comment
Primarily, if you don't worry about dilution you'll loose control (unless you have some sort of special stock that gives you more voting power). Control for me is probably the most valuable thing I have. And, Although selling stock is not the only way to loose control, if you don't worry at all you'll loose control fairly quickly.
Lastly, your employees will also loose moral as they see VC's own pretty much the whole company.
I think Sam knows the Y Combinator investment strategy is best designed to take advantage of power law distributions and getting the very best investments. With crowdfunding and the JOBS act, being able to spread $10k across 50 different companies seems to open the door for everyone to participate in this type of investment strategy. Thoughts?
Of course, you also have be able to identify the good companies, which I will write about later.
How do you resolve the conflict between trying to only invest in good companies (almost regardless of price) versus investing in as many startups as possible (so as to increase your likelihood of investing in a good startup since they are hard to find and identify)?
They stop someone at the street to ask 4 questions. If the person knows the answer, he has to find someone else on the street who also does; if he doesn't know, find someone who doesn't know. Everytime both he and the person he finds either answer correctly or not, the participant gets R$ 250.
Then, after the participant wins R$ 1000, he's offered the chance to find one person on the street who can answer 2 of 4 (trivial) questions correctly to win R$ 2000, or leave with R$ 1000.
The funny thing is that, even when the participants manage to find 4 others who answer all the questions correctly, they refuse the chance to double their money out of fear - even when they proved themselves that the people on the street know the answers!
It's mind-boggling how risk-adverse we are.
By my math, the probability of one given person knowing two answers out of four to win another R$ 1000 is way higher than the probability of four people answering exactly like the participant required to win the first R$ 1000. So it feels intuitive to take the final chance, specially given how much data/options the participant has (knows all the questions and correct answers already, and can choose any geek on the street).
But I'm being purely rational - R$ 1000 on the hand is certainly worth more than a possible R$ 2000. Maybe if the final prize were R$ 10.000 or more people would chose differently?
To win the first prize, four independent events should happen: the participant has to answer (either right or wrong), and a random person in the street has to answer the same way.
Let an event "QnPmAo" mean "Question n, Person m, Answer o", where n=4, and o can be 1 (right) or 0 (wrong). The probability of winning the first prize would be:
P(Win) = (P(Q1P1A1 ∩ Q1P2A1) + P(Q1P1A0 ∩ Q1P2A0)) * (P(Q2P1A1 ∩ Q2P3A0) + P(Q2P1A1 ∩ Q1P3A0)) * ... and so on, for all four questions.
To win the second prize, only two independent events are need: a participant knowing any two out of four questions. The probability would be:
P(Win) = P(Q1 ∩ Q2) + P(Q2 ∩ Q3) + P(Q3 ∩ Q4) + P(Q1 ∩ Q4)
From a statistical viewpoint, the second scenario seems always more likely than the first, assuming the knowledge of the population about trivial questions follows a normal distribution. If during the first round all four participants know the answers, the odds of a fifth participant knowing at least two are high.
Still, from the episodes I watched, people seem to refuse the second challenge regardless of the outcome of the first round - in other words, they don't seem to make a rational decision.
Hope someone less rough than me in statistics can shed some light if there's a mistake!
This is the assumption I question. Knowledge tends to cluster geographically--people in the same area get the same cable channels, on which are aired the same re-runs of Jeopardy on the same nights.
In a nutshell it's this: They now have $1000 in the bank, and worry more about losing that than winning an additional $1000.
What did surprise me was that changing the $5M to $50M still resulted in around 90% taking the $1M, and at no amount could I get a majority of people to choose the random option. At a guaranteed $1M versus zero to $100B, it was split about 50/50.
Apparently even a one thousandth of a percent chance of earning less than $1M is enough for many people to "play it safe". Either that or 50% of people don't understand basic math. ;)
So it's all pretty logical.
I would have chosen random to 5$, but then again I hope I know the tricks how to make money value linear.
Taking 40% as much money (on average) is some really expensive risk mitigation. There are much, much better deals available.
For the larger amounts of money, the risk mitigation comes at a much higher price and is even more silly.
What doesn't, as you say, is considering a random amount between zero and 100 billion dollars in any way risky compared to $1M. You're probably more likely to take the $1M then to be struck by lighting than you are to take the random option and end up with less. (And even THEN, ending up with $900k isn't the end of the world, so the odds are even lower of a really unfortunate outcome.) So you're giving up an absolutely massive expected value in order to avoid a truly miniscule risk. This is risk-aversion at its most ridiculous IMO.
There's no bias left, at least not there.