I think facebook is overrated.

I think everything they create is overrated.

Don't have much love for Facebook as well, but saying something is overrated is poor criticism.

@Makenshi It's not really critique as it has nothing to back it up.

@nblackburn That's true. But what does that has to do with what I just said? Edit: Hey, changing your post is cheating. ; -)

@Makenshi *poof*

Facebook.com may be overrated. But Facebook is not a website, it's a freaking huge infrastructure built and maintained by some of the greatest computer scientists there is.
Seriously, you have no idea what is going on begins the scene (I have no idea either, but each time I hear from them they have built an enormous software engineering masterpiece for a tiny thing in their infrastructure)

@willol Sometimes a huge solution to a tiny problem is the wrong way to look at things.

@nblackburn it's not like they have a choice, they have hundreds of millions of simultaneous requests.
For what's tiny for us (rent a server on OVH), they have to build a whole data center, man it, develop the whole infrastructure, replicate everything, etc etc.
Nothing is tiny at this scale actually

@willol There is choice in everything but it depends how you look at it. The thing about scales is that there is a concept of 'balance' so there size makes the problem significantly smaller because they are better equipped to deal with it.

@nblackburn yup, and how do they achieve this balancing? Freaking huge CS engineering masterpieces. Those companies (Facebook, Google, Amazon...) are the reason we're beginning to have great distributed framework for everything (Kafka, Spark, Openstack, HBase...)

@willol they achieve balance by allocating the right amount of resources to tackle the problem, it's not rocket science, it's common sense.

@willol The largest companies in the world have problems unique to their size. Most companies will never come close to needing that.

@nblackburn ... I'm not going to argue against that. Just knowing the right amount of resources to allocate when and where (while supporting peaks, failures, errors) is far more complicated than everything you and I will do in our lives.

@willol yes because we are one and they are many. Clone yourself a few thousand times and that problem ain't such a problem any more.

Also don't discredit yourself so damn much, you are obviously a smart person and no doubt valuable to this industry (perhaps a little overly passionate about companies that don't give a fuck about you).