My experience echoes the review from Nov 2015. It is pretty bizarre.

The interviewers are pretty set in their solution and not interested in your solution, even though I tried to explain why it is practically better. This does not reflect very well on LinkedIn as an innovative engineering organization; you disregard an opinion just because it is different.

Some examples:

• They insisted on a linked-list-based implementation for a stack, arguing that an array-based one is not optimal during inserts. I already acknowledged it is O(1) amortized and tried to convince them it is optimal in practice (hint: javadoc of ArrayDeque).
• I had to explain to the interviewers that removing items from the end of an ArrayList is O(1), not O(N) like random removal.
• I had to change my generic variable E to T because interviewers thought it was wrong (it does not matter, and E is actually the convention for collections).

And it goes on.