Conversation

1) A telling conversation that just happened in the office: BD: "can we see if there are recent changes in number of FTX deposits?" DEV: "Sure!" ... DEV: "eh this looks like noise, totally dominated by a few large users." BD: "ah right, how about number of users depositing?"

10:18 AM · Oct 24, 2020Twitter Web App

Replying to

2) DEV: "ok cool, I'll get that." It was pretty mundane, and no large conclusions in the end. But there were so many pitfalls avoided there: a) at most places, there wouldn't have been any sanity checking. Dev says "up 30%!" and BD then triples down on a campaign.

SBF

@SBF_FTX

Oct 24, 2020

3) b) the quick back and forth is crucial: this took ~4 rounds before getting to the right query, but only took ~10 minutes. If communication isn't great, it takes a week. c) Totally no pressure to draw a strong conclusion if there isn't one; at a lot of places "meh" isn't ok

SBF

@SBF_FTX

Oct 24, 2020

4) These things seem small but they add up quickly. Often, from the outside, you wonder why companies make bizarre decisions. But the more I dive in to how companies work, the more I'm surprised they ever get things right. A bit of internal dysfunction makes everything break.

OGiberstein.lens (

)

Replying to

and

Great thread

Replying to

and

How refreshing. After reading Dilbert strips for a while, and especially the comments below, one tends to become very cynical. But he's lampooning large corporations.

Adrian Scott

adrianscott.(xch|eth|com)

Replying to

and

The pareto distribution will always assert itself

Replying to

and

What does BD stand for in this instance?

Replying to

and

surprised you don’t have performance dashboards with something as simple as Deposit data etc

Replying to

yeah we totes should we tend to look at volume over deposits b/c deposits are way noisier, but for deposit-related campaigns you gotta do what you gotta do

Show replies