[x0054]

The Jim Keller interview is a wonderful example, listen to the first one. I don't want to re-listen to pull out specific time codes. The first time I listened I was actually interested, I wasn't listening with intent to create time coded indictment of Lex. If you are regular listener and you like his show, I doubt you would ever agree with my take even if I offered you all the proof in the world, because ultimately it's subjective. I say he remains silent and switches topics because he is dumb, but clever enough to be a good bullshit artist. You want to give them the benefit of the doubt.

[zepto]

> I doubt you would ever agree with my take even if I offered you all the proof in the world

You’ll never know if you don’t try.

> I say he remains silent and switches topics because he is dumb

Earlier you said this was obvious. Now it seems like you have no confidence that anyone but you can even tell.

I will listen to some of the Jim Keller episode, but even some clue of where you got the impression from would help.

Edit: I’m 30 minutes in. It’s a great podcast. Nothing obvious about what Lex is doing to make it worse.

[x0054]

This is far from the best example, but here is one:

https://youtu.be/1CSeY10zbqo

That's just what I had a moment to find right now and it was well labeled and I remembered that moment. As for your other comments, my only possible response would be insulting, and I have no desire to insult you.

[zepto]

Also - at 44:50, they have a brief argument about what constitutes ‘search’ in a multidimensional space, and Jim ends up conceding to Lex.

[x0054]

zepto,

Thank you for highlighting that point in the interview. That's another excellent example of Lex being incompetent in the very Science he proclaims to have a PhD in. If you don't know that, you basically just proved why I chose not to get into this argument with you. And Jim didn't concede, he moved on from a really stupid conversation topic onto something else.

[zepto]

> If you don't know that, you basically just proved why I chose not to get into this argument with you

You keep saying things like this, but frankly in so doing you make yourself appear yourself to be intellectually dishonest.

You have been unable to explain or substantiate any of your complaints throughout this conversation, hiding behind the idea that anyone who doesn’t agree with you is too ignorant to be worth explaining it to, which is clearly an evasion.

There is no rational explanation for why you don’t produce an explanation other than that you don’t actually have one.

The simplest explanation is that your hypothesis that you are jealous of Lex is correct, and there is no substance to any of your criticisms, because if there were you’d have presented some.

In case you hadn’t seen it, your posting has been flagged.

[x0054]

So you stated something that wasn't true (Jim conceding), I pointed out that that wasn't the case, and then you flagged my post. So intellectually honest! A true gentlemen!

In any case, let's argue in good faith about this one specific moment you pointed out. The question in hand is if Gradient Descent in ML should be considered a "Search" problem. Lex argues that it is a "Search" problem and Jim is arguing that "Search" has a specific meaning in CS, and ML Training is NOT that.

Jim's understanding of the word "Search" matches my own. Search is a process of looking for a particular data or __specific__ outcome. What happens in ML during training could be described as optimization, filtering, and, obviously, training, but in no way does it fit the definition of "search".

For reference, my definition of "Search" corresponds to the formal definition described hear: https://en.wikipedia.org/wiki/Search_problem, Lex's does NOT, and this is what Jim was pointing out. He eventually gave up when he figured out that Lex was using a colloquial definition of search, not the technical CS definition of the same word, and "conceded" by saying "sure, if that's what you mean by search" or something to that effect.

I would appreciate if you wouldn't flag this comment, but rather engaged with it, like an adult. Thank you.

[zepto]

Yes - that piece came earlier in the episode than 30 mins, so I had already heard it. What is it that you think Fridman doesn’t understand?