Skip to content
Dalton + MichaelDalton + Michael

Should You Keep Working on Your Startup?

Dalton Caldwell and Michael Seibel discuss why you should continue working on your startup. Discussion includes: Running out of hope is why most people give up, the natural state of startups is that they are failing, if it doesn't feel like it's working you are probably right but you can fix it, founders get stuck thinking that things happen quickly, repeating year one, good reasons to quit, why you should stop being your worst critic, copying other people, not talking to users, enjoying the ride, disassemble the "if I only had a job at Anthropic" myth, seeking agency, why you are not alone, why you should find wins, why you shouldn't compare to others, escaping the trough of sorrows, and more. – Standard Capital is the AI-native Series A fund. Learn more at standardcap.com – About Dalton: Dalton Caldwell is Co-Founder and Partner of Standard Capital. He spent 12 years at Y Combinator, where he served as Managing Partner, worked across 25 YC batches, and advised more than 1,000 startups. His investments include Whatnot, Brex, GitLab, PostHog, Retool, Rappi, Razorpay, and Oklo. Before becoming an investor, Dalton founded imeem and App.net. About Michael: Michael Seibel is a Partner Emeritus at Y Combinator, where he served as a group partner and leader of the early stage accelerator from 2014 - 2024. Michael also serves on the board of three companies: Reddit, Dropbox, and Kalshi. He moved to the bay area in 2006, and was a co-founder and CEO of two Y Combinator startups Justin.tv/Twitch (2007 - 2011) and Socialcam (2011 - 2012). In 2012 Socialcam sold to Autodesk Inc. for $60m and in 2014, under the leadership of Emmett Shear (CEO) and Kevin Lin (COO) Twitch sold to Amazon for $970m. – Are you an AI builder? Check out StandardDB. Discover offers, credits, tools, and partner programs from the StandardDB ecosystem.

Michael SeibelhostDalton Caldwellhost
Sep 7, 202614mWatch on YouTube ↗

CHAPTERS

  1. 0:00 – 0:29

    Startups aren’t an EV math game: agency traded for risk

    They open by challenging the idea of evaluating startups purely with expected-value math. The core trade is more personal agency and control in exchange for higher risk and lower predictability.

    • Expected-value cash math often says “don’t do a startup”
    • Founders choose startups for control/agency, not optimal risk-adjusted pay
    • The risk/agency trade isn’t right for everyone
    • If you’re optimizing for stable money, startups are the wrong vehicle
  2. 0:29 – 1:04

    Why most startups die: quitting when hope runs out

    Dalton frames the most common cause of startup death as “suicide” rather than external competition. The episode’s purpose is to re-instill motivation with a rational explanation for why continuing matters.

    • Founders often stop due to lost hope, not inevitable defeat
    • The audience is founders/employees who need fuel and logic
    • Momentum and morale are core survival resources
    • Motivation is treated as a practical founder skill
  3. 1:04 – 1:21

    The simplest logic: quitting guarantees failure

    They underline a basic but powerful point: if you stop, your probability of success becomes zero. Because the default state of startups is “failing,” feeling like things aren’t working is normal—not proof you should quit.

    • Quitting is the only guaranteed way to not succeed
    • The natural state of startups is struggling/failing early
    • Feeling it’s not working is usually accurate—and expected
    • Normalize the discomfort as part of the game
  4. 1:21 – 1:47

    Startup “dysmorphia”: it only looks like everyone else is winning

    They describe how founders misread public signals and assume other companies are thriving. From their vantage point, most companies are also struggling—this perception gap amplifies unnecessary shame and panic.

    • Public narratives create a false sense that peers are succeeding
    • Insiders see that most startups are not going well
    • Comparison distorts judgment and increases quitting risk
    • Social proof via press/Twitter is a misleading dataset
  5. 1:47 – 3:27

    Most successes take longer than you think (and “year one” repeats are a trap)

    Michael argues founders over-index on a few “fast win” examples and underestimate the many long roads to success. A common failure mode is restarting “year one” repeatedly via constant pivots instead of compounding learning.

    • Use AI/research to learn real timelines of successful companies
    • There are both quick wins (e.g., YouTube) and long arcs (e.g., Nvidia)
    • Year one is too early to know if it will work
    • Repeatedly pivoting can prevent reaching real knowledge/product maturity
  6. 3:27 – 4:25

    Valid reasons to stop vs. ego-driven reasons to quit

    They distinguish legitimate shutdown triggers (money/health) from emotional triggers (embarrassment, unmet expectations). The recommended reframe is to give yourself enough “surface area” to get lucky and avoid self-sabotage.

    • Legit stop signals: out of money, health collapse, unsustainable situation
    • Common bad triggers: shame, impatience, “why aren’t we huge yet?”
    • Give the company time to benefit from luck and iteration
    • Don’t become your own harsh authority figure and quit early
  7. 4:25 – 5:20

    The ‘trough of sorrows’ is normal—and success stories hide it

    They name the bleak middle period and emphasize it’s common enough to be widely recognized. Successful founders talk about it only after winning, while investors rarely broadcast failures—so founders lack honest context.

    • The hard middle has a name: “trough of sorrows”
    • You’re not alone if it feels terrible and unclear
    • Retrospective storytelling hides how bad it was mid-journey
    • Investor/founder incentives reduce public honesty about failure
  8. 5:20 – 6:38

    What the trough feels like: thrash, trend-chasing, and not talking to users

    Dalton describes the day-to-day confusion: shifting ideas, bolting on trendy features, and being swayed by tech discourse. This state often coincides with reduced user contact and a search for “authority” to supply a pivot answer.

    • Early-stage work can feel like random idea glue and weekly shifts
    • Tech Twitter/press can drive unhealthy comparison and copying
    • Founders in this mode often stop talking to users
    • Seeking external authority (“tell me what to pivot to”) is usually unhelpful
  9. 6:38 – 7:38

    Reclaiming perspective: being in the game is already a win

    Michael shares how he used perspective—being young and running a company—as a psychological anchor. Even if the company fails, the agency and experience can outweigh the alternative of a long, low-control career path.

    • Running a company can be intrinsically enriching despite bad graphs
    • Agency and decision-making responsibility are valuable rewards
    • Perspective can counter discouragement during slow progress
    • The alternative path often involves years of low autonomy
  10. 7:38 – 10:40

    Debunking ‘you should’ve joined Anthropic’: hindsight and wrong incentives

    They dismantle the common regret narrative that a high-flying job would’ve made more money. It relies on hindsight certainty, questionable hiring assumptions, and an overly narrow definition of success as only cash EV.

    • You couldn’t reliably know which company would explode in advance
    • You might not have been able to get hired at the “winning” company
    • If maximizing low-risk money is the goal, don’t start a startup
    • Hindsight counterfactuals are psychologically toxic and unproductive
  11. 10:40 – 12:13

    Keep going long enough to learn the outcome of your hypothesis

    Michael reframes a startup as an experiment with a hypothesis that takes time to test. The key failure mode is quitting before you’ve gathered clear evidence; pivots are fine when driven by real learning, not impatience.

    • Every startup embeds a testable hypothesis about users/value delivery
    • Time is required to collect decisive information
    • Quitting early can mean never learning whether it could work
    • Pivot based on learned truth, not missed expectations about speed
  12. 12:13 – 12:58

    Tactical mindset: find wins, learn, and value customers—not comparisons

    Dalton offers practical framing to survive: enjoy the process, measure progress by learning and customer value, and avoid the endless misery of comparing to others. Learning in the trough is what enables eventual escape.

    • Look for small wins and evidence of customer value
    • Comparing to others creates lifelong dissatisfaction even after success
    • Track progress via learning, agency, and customer help
    • People who exit the trough are actively learning and shipping
  13. 12:58 – 14:11

    A resilience case study: two years of failure before it finally works

    Dalton recounts a recent investment where nearly everything went wrong—competition, no growth, fundraising struggles, cofounder departure—yet persistence led to success. The takeaway is less about a clever trick and more about endurance through bad luck.

    • Many failure modes can stack at once and still not be terminal
    • Shutting down would’ve been rational—and still would’ve missed the turnaround
    • Persistence can outlast long streaks of “tails” (bad luck)
    • Most success stories contain extended periods of pain before the break
  14. 14:11 – 14:35

    Closing message: success often requires enduring the hard times

    Michael concludes that enduring difficult periods is a prerequisite for reaching good ones, because nearly every successful startup includes prolonged struggle. They end with encouragement to keep going if you can.

    • Hard times are a near-universal component of eventual success
    • If you’re unwilling to endure pain, you likely won’t reach the upside
    • The episode’s bottom line: keep going if finances/health allow
    • Encouragement and sign-off

Get more out of YouTube videos.

High quality summaries for YouTube videos. Accurate transcripts to search & find moments. Powered by ChatGPT & Claude AI.