<aside>
💡 Transcription of email sent by HM to Sifted Opinion Team August 4th 2023
</aside>
Re Atomico (a much more measured take/disproof than yesterday's):
- Ethos: If they were so committed to diversity, they would have done this deep work themselves
- Rationale: It shouldn’t fall to me to define race and explain racism to the most influential and privileged people in the world; there is no reason why they can’t figure this out themselves - I’m supposed to be busy investing just like them (this is a symptom of the industry’s broader mantra ‘strong opinions, weakly held’ - eschewing all individual responsibility and putting the onus on other people to prove them right or wrong)
- Quota vs Target:
- Background: even if this technically may not be a quota, it’s still an explicit diversity target - and they categorically can’t prove what the ratios would be under an unbiased process
- Mathematical Fallacy
- Claim: ’50:50 is the expectation’
- Experiment 1: Baseline
- Assumption: All funds have the same number of people (s)
- Flip a coin for each person (M/F) in the fund, calculate the %M (or %F) and add to a distribution
- Repeat for all funds
- Result: Bell curve - most are equal/near equal M/F split, some are not
- Experiment 2: Fund Size Variability
- Repeat the experiment, but for each fund after each flip continue with probability p, or end flipping for that fund and add the result to the distribution
- Result: likely a Bell Curve with far fatter tails
- Reason: smaller funds are much more likely to have unequal gender distributions, because removing one woman (or man) from a small fund has a far greater effect on the gender balance than removing one woman (or man) from a large fund
- Small Fund: 2:2 —> 2:1, 50% —> 66%
- Large Fund: 50:50 —> 50:49, 50% —> 50.5%
- Further Experiments:
- Investor vs Non-Investor Roles
- Decision Making Investor Roles vs Non-Decision-Making Investor Roles
- Ultimately, this is a multiplicativity fallacy
- Note: this is strictly mathematical, we’d also have to do a neuroscience experiment to confirm or deny any ‘James Damore’-esque claims about whether M or F brains are on average better for the job (i.e. if the peak of the bell curves should/would sit in the middle of each distribution)
- Conclusion: A venture investor should know better than to simply slice off tails from their distributions, they themselves depend on tail distributions for their returns (power law etc)
- Result: either they’re lying about their commitment to diversity, or they’re stupid and don’t understand high school maths, stats, and probability. Given they’ve invested in quantum computing companies like PsiQuantum, which rely on maths, stats, and probability (noting also that they have at least one dedicated expert on this in house) - I find it very hard to believe that it’s the latter.
- Hence, we can demonstrate that they aren’t committed enough to driving diversity in vc for the amount of reach that they have (this applies to all of the funds, not just them)