Garvit Sahdev

Breaking Next Big Ideas at 'The Thoughtful Tangles'

[Arindam Paul @arindam___paul]

This is very well written

[Garvit Sahdev @garvitsahdev]

@Manu_Sisti @NavalismHQ Claude

[Dharmesh Ba @dharmeshba]

Ask any engineer why they prefer quantitative data and they will give you a dozen reasons.

All of them are wrong, or at least incomplete.

The actual reason is much simpler, and much more human: numbers let you cheat.

You can take a mediocre finding and rescue it with a formula. You can redraw the baseline. You can change the time window. You can present the same flat result three different ways and, with enough effort, make it look like a trend. The entire field of dashboard design is, quietly, the study of how to torture a number until it confesses to the story you wanted it to tell.

Stories don't permit this.

When a user looks at your onboarding and says "I felt stupid," there is no pivot table that saves you. The quote sits there, and either you face it or you don't.

This is why quantitative research dominates in organisations where careers depend on being right. Not because it's more accurate - it is routinely less accurate than a single good interview - but because it's more defensible.

A PM who presents a number that turns out wrong was let down by the data. A PM who presents a story that turns out wrong was let down by their own judgment.

The first carries no personal risk. The second carries all of it.

This is why the companies that genuinely understand their customers tend to be run by founders who are, by temperament, slightly allergic to spreadsheets.

They have, whether they know it or not, opted into a world where they can't hide.

Which is also, not coincidentally, the only place where real insight lives.

[dharmesh @dharmesh]

AI is getting better and better at answering CAPTCHAs.

Me? I'm failing them more often.

Soon, I'm going to have to start using AI to prove that I'm human.

[Dharmesh Ba @dharmeshba]

Most founders believe product-market fit is a question about the product.

It isn't.

It is a question about the customer's priority list - an ranking they carry around in their head, mostly unconsciously, of the things currently bothering them enough to act on. Your product is not competing with other products in your category. It is competing with whatever is ranked number one through six on that list.

If your problem is ranked seventh, you do not have PMF. You have an interesting thing the customer will talk about in a focus group and never pay for.

The test is not whether the problem exists. Almost every imagined problem exists, in some diluted form, somewhere. The test is whether the person has already done something about it. Have they Googled it at midnight? Have they taken a class? Bought a course? Paid for an inferior solution? Complained about it to three friends this month? If the answer is no, the problem is not top-of-mind. It is merely real.

Real problems are cheap. Top-of-mind problems are the only kind anyone will pay to solve.

This is why so many launches fail despite excellent research, brilliant design, and accurate diagnosis of a genuine gap in the market. The gap was real. The gap was not urgent. The customer nodded along in every interview and then, back at their life, continued to ignore it.

The best founders don't test whether their product is good.

They test whether the problem is ranked high enough for anyone to change a behaviour over it.

[Paras Chopra @paraschopra]

Pragmatist philosophy is basically a cheat code for understanding.

What it says is that understanding something is to build a model of it. What happens is that we build models of reality and adopt those that serve us well (in terms of usefulness or predictive accuracy) and reject those that seem wrong or arbitrary.

We can’t access how reality “really” is as we can never peer beyond the concepts we have. Reality is parsed through our theories and models. Although, we can iterate on our concepts (which is the whole project of Science).

So, if a metaphysical concept has no empirical implication on the world, don’t bother. If
it has, include it in your model.

If you internalize this, many confusing questions become actionable.

“Do we have soul?” -> what do we mean by the word “soul” and how do we observe whatever we mean by it? Is there a simpler explanation for that observation?

“Do AIs have feelings”’-> do we have a model for feelings that we can extrapolate to AIs? Why should we believe in that model vs simpler model that they’re faking emotions?

“What should I do in life” -> what’s a useful model of a good life that’s applicable to you? What’s missing between how things are and what the model says.

The list goes on..

[Paras Chopra @paraschopra]

The most useful way to think about AI capabilities is to think of target domains in terms of 3 orthogonal axis:

- Ease of building a verifier (coding is easy, lab equipment manipulation is hard)

- Causal complexity in terms of number of confounds (math problems is low as answers don’t depend on external factors, startup success is high complexity as it depends on many random factors)

- Economic attractiveness (high for coding, low for many domains)

What we’ve seen the first to be automated are domains where building verifiers is easy, causal complexity is less and economic attractiveness is high.

No wonder coding is the first one to see big jump.

But inflated valuations of AI companies need to be justified, so I fully expect that these companies will keep attacking the next best domain they can until they exhaust the economic attractiveness constraint.

[Raj Shamani @rajshamani]

The loudest person in a negotiation usually gets the worst outcome. Volume signals need, and need is the only information the other side requires.

[Varun Mayya @waitin4agi_]

Humanity has run on a thin violence minimization program in the form of allowing anyone to use their life's energy to earn upward social mobility. Remove that and we're back to the jungle lol.

[Garry Tan @garrytan]

AI means people need to stop thinking so much about what their boss wants

They need to start thinking much much more about what customers want

And that will make a big difference in society: too much of it is caring about Keynesian beauty contests and other such contests that are disconnected from the direct needs of others

Less dead weight loss. More actually helping people.

[Garvit Sahdev @garvitsahdev]

@naval Action filled life is so relieving :)

[Garvit Sahdev @garvitsahdev]

Don't believe in what you are selling. Sell what you believe in.

[Garvit Sahdev @garvitsahdev]

@amrishrau Amazing :)

[Amrish Rau @amrishrau]

We got a patent for a revolutionary idea for Online payments.
Tap card to your own phone to complete Online payment transaction while sitting at home. Have a look.

[Pratyush Rai @pratyush_r8]

We all know deep down what’s real and what’s BS.

Though we still end up trusting our own BS & consider it to be a productive path to get the outcomes we desire (which never happens).

Having the ability to watch out for our own BS and cut it off is nothing short of magic.

[Dharmesh Ba @dharmeshba]

Most Indian consumers don’t articulate needs.
They adapt to pain.
Your job is to observe friction, not wait for feature requests.

[Paras Chopra @paraschopra]

The human brain is a compression machine.

If your days are too similar, your mind will compress them into a single memory.

If you want a long life, live a varied one.

[Garvit Sahdev @garvitsahdev]

https://t.co/ubhfvko5r5

[Garvit Sahdev @garvitsahdev]

@kunalb11 Aahaa