Real Alpha
Watch mode

Training Data · 5 min instead of 1 h 6 m · Oct 8, 2026

Hiring signals: Harvard predicts little, interviews must be scripted

LF

Luca Ferrari

Co-founder & CEO, Bending Spoons

  • Sonya HuangHost · Partner, Sequoia Capital

    Works with OpenAI and Anthropic; co-wrote Sequoia's essay on generative AI.

  • Pat GradyHost · Co-steward, Sequoia Capital

    Joined Sequoia in 2007; backed Snowflake, Zoom, ServiceNow and OpenAI.

About this segment

Luca explains which application signals Bending Spoons found predictive, including a surprising one about teenage politics. He also says unscripted interviews are noise and shows how to structure them.

From “Luca Ferrari: The Operating System Behind Bending Spoons”, Oct 8, 2026. Watch the full episode

“We found that if someone is politically active as a teenager, but CZs to be in their early 20s that tends to correlate highly with someone's drive”
LFLuca Ferrariat 14:57 · Training Data14:57

Takeaways

  1. Harvard or Stanford adds little predictive power and can breed entitlement
  2. Teen political activity that stops by the early 20s signals drive
  3. Scripted, separately scored interviews beat gut-feel chats

In this episode

Full episode
  1. 13:28Hiring signals: Harvard predicts little, interviews must be scripted4:54

  2. 31:59Why Bending Spoons favors talent and motivation over experience3:23Up next
  3. 55:09Make whoever adds complexity carry the burden of proof1:39

Transcript

Transcribed by machine (Whisper) from the episode’s audio, so a word may be off here and there.

  1. 13:40

    What are the signals that are most strongly correlated with success? The R-squared's high. And then once you get beyond that, I assume there's a human interview. What are the questions that tease out what you're looking for that maybe aren't obvious? Yeah, so we're fairly secretive about the details. I guess you get a similar answer from my quantitative trader. That was my metaphor earlier. But now I can share some of this.

  2. 14:13

    There are some obvious signals in people's applications. Their GPA, for example, is not super strong, but it's helpful. the university graduated from, I think people will be surprised by how poorly graduating from Harvard or Stanford predicts performance. Like people think, oh, you must be a genius. No, actually, yeah, you're probably a little bit stronger than, you know, people graduating in other universities, but not so much stronger.

  3. 14:42

    And often we find that people tend to lose some of that potential raw talent edge by being a bit more entitled and thinking too highly of themselves. But it's still predictive. I mean, all else being equal, someone from Harvard will do a little bit better than someone from a less well-known university. So some of those are probably somewhat obvious. There are many that are not. One I can share, which is relatively predictive.

  4. 15:07

    I was surprised when this came up. We found that if someone is politically active as a teenager, but CZs to be in their early 20s that tends to correlate highly with someone's drive and entrepreneurial mindset and the way we have explained it which at least are highly valued qualities the way we have explained it is that if you are 17 years old and school is not enough you want to, you feel like you need to do more, that shows agency, that shows productivity, that shows a level of not settling for the status quo that's generally healthy. But if you still do that when you're 25, 26, 30, it means it's probably a life mission, which is perfectly fine, but likely incompatible with your job, your career being, you know, top priority. Your job is probably probably takes a backseat to your political aspiration. How do you find that signal?

  5. 16:11

    Do you ask them that question in an interview? Do you look through their old Instagram? Like, how does that signal even arise in the process? Thankfully, people think they need to include leadership experiences in their CVs because that gets you a job at, you know, a consulting company or something. So most people we find would include something significant in that area if they did it. But in any case, we do ask later in the process, whether it's an interview or written assessments, questions that would very clearly lend themselves to people sharing this sort of thing.

  6. 16:43

    And with this, I'll get to the other thing you can do, which is you shouldn't be passive in accepting whatever signals are present in someone's application. You can be very proactive. we find that a lot of the best signals in our case come from the practical tests, the things we do or ask candidates to go through. And so you can elicit signals that way. Interviews are, of all the typical elements in a selection process, at least in our experience, are the worst predictors.

  7. 17:19

    but you can make them decent if you give interviewers extremely precise questions they need to ask, and you tell them to ask them exactly in the same way. And then you make sure that the answers are recorded and assessed separately. So what's completely useless, pretty much, is having someone talk to someone else with essentially no guidelines or just a high-level agenda, and then tell you, I like the guy, I didn't like the guy.

  8. 17:47

    That's noise, pretty much. So if you structure it a lot and you try to separate the actual conducting of the interview from the evaluation of the responses, then it can be somewhat predictive. Can I tell you about HubSpot unlocked on the sales side a good process for doing this where we scaled our inside sales org from 10 to 1,000?