I'm not sure I'd say "New Delicious was rewritten in Erlang" is an accurate title here. More like "certain subsystems of Delicious were rewritten in Erlang." As far as I know no one (or at least hardly anyone) is actually writing web-applications themselves in Erlang, they're picking off the subsystems that fit the language's sweet spot.
Not that this project isn't cool, but it seems like the ICFP must be desperate for presentations if this is what suffices.
When we were doing Powerset's infrastructure I didn't even think of writing a whitepaper or anything, I didn't realize that "Just doing it functionally and in a distributed fashion" was grounds for attention.
I'm familiar with some of the strengths of Erlang, but the slides look like they're written from the point of view of someone that likes Erlang and was looking for a way to justify rewriting Delicious in it, rather than explaining what about Delicious made Erlang a wise choice for the task.
someone that likes Erlang ... what made Erlang a wise choice for the task
You've just answered your own question. You can solve every problem in every programming language -- as long as you like what you're doing. If you pick a langauge you like and a project you like, it will probably get done.
The presentation is definitely worth a look. Evidence of erlang becoming a solution to real problems is coming up more and more, and the examples (like this) are significant.
unless its php, java, or perl, (in order of precedence), it is not a web production language at yahoo
and beyond fanboying, there is nothing particular about erlang that makes it appropriate for this problem. and don't say "scaling"...yahoo's answer to scaling is the same as google's...lots of machines (far more than needed), and lots of money.
The script that copies the data from the old system to the new one is what they wrote in Erlang.