Ahmed El-Kishky

And Jakob was a very accomplished

You know, competitive programmer at ICPC, at Google Code Jam, winning so many.

So they were always kind of our goal.

And so we sort of made a chart.

The chart itself was sort of like the ELO score of our model.

And it was like 300 or 400 ELO when we started out.

And we had this line that was just a Mark Chen line, but it was like above 2000 ELO.

We had a nice little picture of Mark Chen.

And then we had a line a little bit above that of Jakob Oczaki at like 2600 ELO.

And it seemed so far away.

And basically, both Mark and Jakob would like, you know, encourage us because they were excited by, you know, these models getting better because they knew the implication there of having a powerful model.

And like over the year or so that we were like, you know, trying to see how good our models are, it would just like inch up a little bit on that chart.

So like, oh, this model is, you know, 600 ELO.

This one is 1,000, 1,500.

Until one day I remember messaging Mark to be like, I have some bad news.

And it was...

we saw that like that model, I just like slightly edged, uh, and just, uh, craft above Mark's ELO score on code forces.

You sir have fallen.

That's amazing.

Like not even like maybe a month, month and a half later, um, Jakob, you know, was also slightly beaten and, um, yeah, they basically were incredibly excited because they knew the implication.