But it's important to get a large sample size

Maybe. It certainly isnt some inviolable principle. Maybe it is useful.

10 mil will still give you a more accurate result than 100k

It doesnt make it MORE TRUE

10 mil vs. 100k in coinflips

It's not about being "more true" but more accurate

Will not give you a more accurate result

it will

More accurate what though

Than what you'd likely otherwise get just doing 100k

in other words

Hang on

sure thing

100k fully establishes a certain premise

it does

and that premise is valid

How can you get more accurate than 50 50

why do you assume you'll get a perfect 50/50 at 100k?

Will it become 50.0n1 on one side?

You likely won't

but you'll get close

You will.

and you'll get closer at 10 mil

Why stop at 10 mil

You believe someone will always flip exactly 50k of their 100k flips on heads every time they attempt it?

Go for broke

you can stop at whenever you want

10 mil study under the same circumstances will be appreciated more than the 100k

and there's valid reason for it

No. But you can nevertheless extract the trend at 100k. 10 mil helps you not one bit.

Not even a little bit

Well it gets you a more accurate result

Its ridiculous you think so

Whether you think it's accurate enough is subjective

This is exactly what I meant

I'd certainly trust a 10k study over 10, and a 10mil study over 10k, though with all things being equal

statistics are ridiculous?

Obv not as stats is stats at 100k or 10 mil.

Stats is not amounts

well, again you're free to think what you want

Yes, we all have budget and time constraints