← Back to list
Colossus (Invest Like the Best / Business Breakdowns)Podcast1 Dec 2020Source: traffic.libsyn.comHost: Patrick O'Shaughnessy

Daniel Gross – Finding Undiscovered Talent - [Invest Like the Best, EP.202]

In plain words

Daniel Gross says traditional personality tests like MBTI are unreliable because they lack market validation. He built Pioneer, a platform that uses a real-time scoring system (like a 'PageRank for people') to find persistent, insecurity-driven talent. He's bullish on Starlink, citing 1000x lower launch costs and near-impossible regulation. He's neutral on GPT-3: good at short text but loses focus on deep analysis. TikTok faces geopolitical risks and may need to divest its US operations.

AI SummaryAI-generated · may contain errors · verify against the original

Daniel Gross discusses in the program how to discover undiscovered talent through Pioneer, a remote startup generator. The core argument is that identifying talent should not rely on psychometrics, as its scientific validity is limited; instead, more attention should be paid to persistence and "inse

~12 min full read · 9 sections
Deep Analysis

Daniel Gross – Finding Undiscovered Talent - [Invest Like the Best, EP.202]

Episode Overview

Daniel Gross is the founder of Pioneer, a company positioned as a "fully remote startup generator" that uses software to screen and incentivize potential entrepreneurs worldwide. This episode explores the core thesis: talent identification should not rely on psychometrics, but rather on quantitative metrics of actual progress and the personality trait of 'insecurity-driven achievers' — Gross argues that psychometrics lacks scientific validity due to the absence of free market pressure, and Pioneer is attempting to build a real-time scoring system akin to "PageRank for people" to replace traditional screening.


Theme 1: Psychometrics is Overrated – Science Lacking Free Market Validation is Unreliable

Daniel Gross argues that psychometrics (such as Myers-Briggs, Big Five) are popular but have a weak scientific foundation, primarily because they lack free market pressure to validate their effectiveness.

Gross points out that psychometrics attempts to reduce "people" – an extremely high-dimensional object – into numbers, inevitably losing information in the process. The core question is: Is this "lossy encoding" as imperceptible as an MP3, or does it lose key pixels like an over-compressed image?

Comparison Dimension Psychometrics Weather Forecasting
Pressure Mechanism Lacks free market testing Continuously validated by free market
Perceived Validity Effects are exaggerated Limitations are clearly recognized
Typical Example In Big Five, the R² between IQ and income is only about 0.6, far from explaining most variance People understand the boundaries of its accuracy

Gross particularly emphasizes that Myers-Briggs, in his view, is "extremely inaccurate and does not regress to anything substantive." Even the most respected Big Five personality scale has "very light" effect sizes. He quotes his sister (a psychologist), calling it the "Big Five halo" – exciting at first, but the effect fades after a year.

Key Reason: Psychometrics is mainly based on self-assessment, and self-assessment fluctuates significantly over time and emotional states. Gross uses himself as an example: "My assessment of myself at 8 a.m. after drinking coffee is completely different from my assessment at 2 a.m."

Reader's Note: As the founder of Pioneer, Gross's perspective carries a promotional stance for his software screening system – he needs to argue that "traditional methods don't work, my method is better." This is a position-holder's viewpoint.


Theme 2: 'Persistence' and 'Insecure Overachievers' – The Underlying Logic of Identifying Talent

Daniel Gross argues that the most critical indicator for identifying talent is not talent itself, but persistence – the ability of an individual to 'take multiple shots,' which allows the compound effect of time to work on the person.

In creative professions (such as software engineering and investing), the gap between the top performer and the median can be 10 times or even greater (John Carmack is 10 times better than Gross himself). Therefore, the goal is to find those who 'treat work as a performance opportunity and are willing to do as much as possible.'

'Insecure overachiever' : Gross believes that insecurity is like 'molecules seeking covalent bonds,' driving people to find objects for which they can perform, thereby generating enormous energy. He uses Whiplash as a screening tool – this film can distinguish between those who have a confrontational relationship with work and those who have an inspirational relationship with work.

Key mechanisms:

  • Entrepreneurs are 'excellent at maximizing hierarchies but poor at choosing hierarchies' – they will drive any given score with great intensity, but may not necessarily choose the right target.
  • Capital is a good optimization target because it is naturally scarce; while social media likes/retweets are a bad optimization target because they are infinite and not scarce.
  • The success of immigrants partly stems from 'insecurity': being uprooted from their original hierarchy and placed into an entirely new environment; this insecurity drives outstanding achievement.

Falsification condition : If a person is too complacent and feels too comfortable within their own hierarchy, they may lose the source of this drive.


Theme 3: GPT-3 — Dual Lens of Tech Hype and Commercialization

Daniel Gross argues that GPT-3 is "a significant advance in the context of a shortage in the tech world," but whether its commercial application is disruptive innovation or sustaining innovation remains unclear.

The essence of GPT-3: a machine learning model that has made notable progress in natural language understanding, capable of writing articles, summarizing text, and answering questions. Gross compares it to a model with "attention deficit" — it excels at handling shallow text (e.g., tweets, a paragraph or two) but "loses control" when deep understanding is required.

Analysis of the hype reasons:

1. Shortage of technological progress: Since the iPhone, the world has lacked truly major new technological breakthroughs; GPT-3 fills this gap.

2. Ease of experimentation: OpenAI released an interactive website that allows users to experience it directly. "4 out of 10 attempts yield good results, but people pick the best ones and post them on Twitter."

3. It is itself a "meme" — Gross's analytical framework holds that any technology frenzy should be deconstructed as "what gap in the world does it fill."

Uncertainty of commercial prospects:

  • If GPT-3 is a disruptive innovation, startups could use it to create a clone of Shopify (just by saying "build me a website").
  • If GPT-3 is a sustaining innovation, large tech companies (e.g., Shopify itself) would directly integrate the feature, giving startups no advantage.
  • Key question: Training costs are on the order of several million dollars. In the future, almost all large companies may train their own similar models, and the competitive landscape will evolve rapidly.

Theme 4: Starlink and Satellites – From "Technological Frontier" to "Unregulable Public Utility"

Daniel Gross argues that Starlink represents a rare technological form—it is transitioning from a "frontier technology" to a "nearly unregulable public utility," creating enormous investment opportunities.

Why unregulable? Starlink puts internet signals "in the air," requiring no ground towers or cable laying. Although it could theoretically be restricted by banning the sale of receivers, the convenience factor makes full regulation extremely difficult (by analogy with China's firewall, which, though not completely sealed, is already effective enough).

Magnitude of the cost revolution:

  • Traditional Iridium satellite launch: about $50 million per launch
  • SpaceX Starship launch: about $50,000 per launch
  • A 1,000-fold cost reduction will fundamentally change "what can be sent into space"

Non-consumption disruption: Gross cites Christensen's framework, arguing that the most interesting disruption often comes from entirely new use cases, rather than cannibalizing existing markets. Just as when mobile phones were introduced, no one could have predicted Uber—it expanded the total addressable market (TAM) rather than merely replacing taxis. For satellites, the internet is the most obvious initial use, but over the long term, "weird stuff will grow out of it."

Investment timeline: Gross judges that the 2020s (2020-2030) will be "the decade of Starlink and satellites," but investors need to distinguish whether this is infrastructure investment or venture capital—the two have completely different return cycles and risk profiles.


Theme 5: "Seed vs Leech" Ratio — Relationship Capital in Venture Capital

Daniel Gross proposes the "seed vs leech ratio" as a framework for measuring the quality of relationships in venture capital, drawing an analogy to the share ratio in BitTorrent communities.

Core of the Framework: In venture capital, early-stage rounds (Series A/B) are an "access game" rather than a "picker's game" — the good companies are known, and the key is whether you can get an invitation to participate. At this stage, investors need to build "seed-type" relationships: when others meet you, you need nothing and simply provide value.

Quantification Method: Gross suggests reviewing your own WhatsApp or iMessage chat history — who are the people you almost always invite to the "party"? What traits do they have? He wants to be the kind of person who "gets invited" rather than someone who "always needs to ask."

Relationship to Investing: For public market investors, this framework is less applicable (since they can buy anything at market prices), but for private market investments that require access, it is crucial.


Mentioned Targets

Target Guest Attitude Key Data
GPT-3 (OpenAI) Neutral/Observing Single training cost on the order of millions of USD; about 4 out of 10 attempts yield good results
Starlink (SpaceX) Bullish Starship launch cost approximately $50,000 per launch (vs. Iridium's $50 million); almost unregulatable
TikTok Neutral/Event-driven Banned entirely in India, banned by the US military; needs to be spun off into a US entity
Facebook Risk Warning Transitioning from "technology" to "public utility," facing regulatory risk
Twitter Risk Warning Essentially a media company, facing content regulation
Robinhood Neutral Drove the surrounding "shovels and tools" ecosystem
Matrix Neutral Open-source chat ecosystem, spawning multiple startups
Shopify Not specified Could be disrupted by GPT-3 or integrate the feature itself
SpaceX (Starship) Bullish Cost reduced by 1,000 times, making satellite internet a reality

Judgments Worth Remembering

1. "Psychometrics lacks free-market pressure, so its science is unsound" — Gross argues that without market testing, the validity of psychology and epidemiology is systematically overestimated; weather forecasting, by contrast, is tested by the market, so its limitations are accurately understood.

2. "Entrepreneurs are excellent maximizers of hierarchies, but poor choosers of which hierarchy to join" — They will relentlessly drive any score they are given, but may not select the right target; capital is a good optimization target (because it is scarce), while social-media likes are a bad one (because they are infinite).

3. "Pioneer's scoring system is like 'PageRank for people' — a score you can actively drive" — After registering, a user receives a score representing the combined quality of "you + your project," which can be actively improved each week, similar to a video game.

4. "GPT-3 has an attention deficit — it handles shallow text well, but loses control when deep understanding is required" — This analogy explains why GPT-3 is great at writing tweets but "goes off the rails" in deep analysis; it is also a key limitation for its commercialization.

5. "Starlink is almost impossible to regulate — because the internet signal has been 'put into the air.'" — Gross calls this "the most under-appreciated frontier market of the 2020s," but investors need to distinguish whether it is an infrastructure investment or a venture-capital investment, as the return cycles are completely different.

6. "Seed vs. parasite ratio: a framework for measuring relationship quality" — An analogy to BitTorrent share ratios. In access games, aim to be "the person who is always invited," not "the person who always needs to ask." Gross suggests using iMessage chat logs to examine yourself.

7. "Truth is only three questions away: keep asking the same question until you get a truthful answer" — Gross believes that the "rationalized explanations" people give are almost never true; repeated questioning can penetrate performative answers and reach true motives.

8. "The Power of 10 Playbook: everything great starts small" — Gross advises entrepreneurs to first attract 10 users who genuinely love the product, rather than initially aiming to "become Stripe or Airbnb." He cites SpaceX's origin story: it began as nothing more than a PR stunt to "take a picture of a green plant on Mars."