Contrasting Sam Altman and Paul Graham's ideas on doing great work, with LLMs

July 13, 2023 at 03:22 PM

Paul Graham’s recent essay How to Do Great Work was in my opinion one of his all-time best. It also reminded me Sam Altman’s How to Be Successful.

On the surface, they seem to talk about fundamentally different things - outcome vs process, with Altman perhaps focusing a little more on material success.

But they are both talking about how to approach your work, with Altman explicitly saying

I’ve observed thousands of founders and thought a lot about what it takes to make a huge amount of money or to create something important. Usually, people start off wanting the former and end up wanting the latter. Here are 13 thoughts about how to achieve such outlier success.

Short of a new interview with both, this is their most distilled pespectives on doing important work that is available in public.

I thought it had interesting parallels with the evolution of YC as well, but so as to keep my opinions out of it, I had a language model compare them. It was a good chance to take Claude2 for a spin. With its 100k context window, it could fit the full text of both essays easily. At 19k tokens, it also fit into GPT4-32k, which I offer as a comparison.

Table of contents

Open Table of contents

Claude’s comparison

Here are the key similarities and differences between the essays by Paul Graham and Sam Altman:

Similarities:

Differences:

In summary, while both share a growth mindset and urge ambitious projects, Altman is more focused on entrepreneurial tips for career advancement, whereas Graham provides philosophical advice for producing meaningful work.

GPT-4’s comparison

Similarities between Paul Graham and Sam Altman:

Differences between Paul Graham and Sam Altman:

Expert Analysis:

Both authors recognize hard work, long-term vision, and personal interest as indispensable ingredients for success. However, their paths diverge considerably when looking at the roles of self-belief, risk-taking, networking, and curiosity.

Altman’s focus on wealth creation, belief in oneself to the point of delusion, and the value he puts on risk-taking, distinctly reflect how he perceives success. He convincingly argues, “Self-belief is immensely powerful. The most successful people I know believe in themselves almost to the point of delusion”. Such views highlight his opinion that self-confidence is an essential quality in one’s pursuit of success and affluence.

In contrast, Graham values curiosity and continuous learning as a critical path to success. He posits that a person’s curiosity always leads them to the right line of work and sustains them in their journey of achieving greatness. Graham states that “Curiosity is the key to all four steps in doing great work”. He also doesn’t emphasize the need to take high risks, nor does he encourage networking as a key to success.

Despite these differences, both authors provide valuable perspectives on what it takes to do great work and be successful, thus offering valuable insights for individuals in various fields.

Claude Sample 2 (re-sampling often captures new points)

Here are the key similarities and differences between the essays by Paul Graham and Sam Altman:

Similarities:

Differences:

In summary, while both share a drive to do impactful work, Altman’s worldview aligns more with entrepreneurship and business, while Graham has a perspective anchored in the arts, sciences, and intellectual pursuits.

GPT-4 Sample 2

Similarities in the worldviews of Paul Graham and Sam Altman:

Differences in the worldview of Paul Graham and Sam Altman:

In-depth Analysis: Throughout Altman’s essay, he seems to advise readers to seek exponential growth and wealth generation, emphasizing on compounding oneself: “A medium-sized business that grows 50% in value every year becomes huge in a very short amount of time” (Altman). Conversely, Graham is more focused on the intrinsic rewards of doing great work and personally evolving. Altman suggests that a baseline degree of success can be achieved through privilege or initial effort. Yet, he highlights that success is sustained by seeking interesting projects, continuously learning, and patiently waiting for exponential results.

A crucial point of divergence between Altman and Graham is their perspective on risk. Altman encourages early-career risk-taking as the best time for learning. He sees these calculated bets as steps towards outlier success and advocates for taking larger bets in favorable directions. On the other hand, Graham suggests a more cautious approach.

Both authors agree on the importance of having good colleagues and networking with talented people. However, Altman suggests a more strategic approach, focusing on adding zeroes to one’s success metric, while Graham advocates for working with people whom one admires and wants to be like, stressing the importance of mutual growth rather than personal gain.

In conclusion, both authors provide insightful, albeit slightly different, perspectives on achieving success and doing great work. While Altman focuses more on financial success and strategic moves, Graham places more importance on intrinsic motivation, passion, and patience. These variances not only reflect their personal experiences but also underline that there are multiple routes to success.

Source Code

prompt = f"""The following are two essays by Paul Graham (How to Do Great Work) and Sam Altman (How To Be Successful).
Create a bullet point list of ways in which the writers' worldviews are similar.
Create another bullet point list of the ways in which they think differently from each other.
Write an expert level analysis. Be detailed and use quotes to illustrate similarities and differences.

=== START Essay 1: How To Be Successful, by Sam Altman ===

{content_sama}

=== END Essay 1 ===

=== START Essay 2: How to do Great Work, by Paul Graham ===

{content_pg}

=== END Essay 1 ===


"""

claude_completion = anthropic.completions.create(
    model="claude-2",
    max_tokens_to_sample=10000,
    prompt=f"{HUMAN_PROMPT} {prompt} {AI_PROMPT}",
    temperature=0.2
)

gpt4_completion = openai.ChatCompletion.create(model="gpt-4-32k-0613", messages=[{"role": "user", "content": prompt}])

Notes and Observations