Could it be possible to fabricate a presumably 2 years old blockchain in less time (or even instantly) ?

Recently a coin has appeared that claims to be launched two years ago, however, no evidence can be find about it but the blockchain with the actual timestamps.
So I am asking, is it possible to manufacture a blockchain that appears 2 years old but has been recently forged ? The chain has the difficulty and cumulative difficulty values, so I gues this could be possibly checked.

The blockchain in question uses the same concept as Bitcoin (the same byzantine generals solution) but with a different implementation and different PoW.

Why are you beating all around the bush?  Which alt are you talking about?

What coin are you talking about?

I think the difficulty can't be faked, you would need time to compute.

he refers to bytecoin

In the first place, does anyone you know (I mean, REALLY know, as in know their real name/ID and could potentially take them to court if they were scamming you) actually know of this coin before the recent appearance?  If not, I would assume it's a fake and stay away.

In the second place, yes, you can in fact do that.  People launching an attack on existing altcoins in fact use this technique on a frequent basis when preparing 51% attacks with block chains that aim to create a fork where they get the block rewards that the real chain has already awarded someone else.

Here are the steps.

First they take their client source code and replace the calls to get the system time with calls to a shim that returns a fake time stamp.  This shim returns a time that starts a couple years ago and advances 100x as fast until it gets up to the "real" time.  This will affect what the client will write into blocks as timestamps, AND also the difficulty adjustment it will apply between blocks in order to adjust block timing.  So, if you're solving a chain with ten-minute blocks, as in Bitcoin, your fiddled client would be writing time stamps approximately ten minutes apart, and adjusting difficulty to get such blocks about once every six seconds.

Then you set up two of these fiddled clients and let them talk to each other and create blocks.

A week later, you have a two-year-old block chain.  But the cumulative hashing power (its "length" when we talk about the "longest chain") - only has a weeks worth of hashing from the two rigs you sicced on it.  If you have a couple of 20GH/s Antminers, then it'll look like your chain was sucking up about 400KH/s of hashing power - trivial compared to Bitcoin, but believable for a very small alt.

Now you go back, remove the timestamp fiddling from your client source code, and release - keeping, of course, your newly created 2-year premine. 

So, the question is whether the blockchain you are looking at is or is not fake.  So look at it, make a graph of how much hashing power it was getting over time, and see if the graph looks "realistic", meaning does it look like the hash power graphs of other coins that were running at that time? 

yes, imagine you going to update your block.file and I can sent it on Dark Fiber Channel without traffic, no delay etc ..

I do recall an alt called bytecoin that has been around for a while.  But it appears that there are two different bytecoins, which adds to the confusion.  gmaxwell has said that the "cryptographically interesting bytecoin" has some interesting privacy features (and also some issues with it). 

I do not know if the "cryptographically interesting" one is the same as the one that has been around for a while or not.

btw, "dark fiber" jargon in networking is the cable that was rolled out but is not in use. It's makes sense ... if you going to rool out cables (layer 1) you should rool an extra one as a spare cable. Sometimes networking engineers use theses cables to perform tests ... like latency measurement, etc since this cable have no traffic.

http://en.wikipedia.org/wiki/Dark_fibre

It's very hard to fabricate such a thing but it is still possible. Hope it will not happen