[g-mork]

Just for sheer geekery's sake probably the ISDN talk.

For OMG eye opening factor the FreeBSD jails talk (how the hell is this thing still so buggy?) and the talk on unencrypted satellite links

For excellent follow-along value and dedication to ridiculously pointless cause the Freebox talk. "Technically I don't own this box so instead of risking damaging it I'm going to take the extremely long and entertaining route around, somehow involving Doom WAD files"

For showmanship probably the Tegra talk

[jacquesm]

> For OMG eye opening factor the FreeBSD jails talk (how the hell is this thing still so buggy?)

Because everything that complex is going to be that buggy.

With the bugs they found fix a constant number of them still remains.

[pantalaimon]

Linus said 'many eyes make all bugs shallow', but compared to Linux, there are not many eyes looking at FreeBSD.

[jacquesm]

Linus has said a lot of stuff over the years and not all of it was on the money. Still, he did a lot of good and I'm very grateful for it, Linux has been my daily driver for almost two decades now (basically from when I stopped using SGI because there was no point any more).

But bugs in large codebases will always be a thing, and even though the eyes looking at FreeBSD are very, very good eyes, indeed there are not enough of them. The more interesting thing here is that they picked a really hard target. If they had done the same with Linux I would expect the number of bugs to be quite a bit higher.

[craftkiller]

That "many eyes" theory has failed us many times. For example, OpenSSL's heartbleed or the recent React RCEs.

[ahartmetz]

”Most bugs are shallow” is more like it. One could also argue about the number of eyes actually looking at certain parts.