Retroviruses don't exchange genetic material from other individuals into their host. It's the host's genetic material that is altered, not the viruses.
This is the evidence that it is possible to move information from outside of DNA into DNA. And if it is possible then it can be done by something else than retrovirus. It is enough that such process took place very rarely with the frequency comparable to the mutation to change completely the dynamics of evolution.
Such reverse update could be nonrandom and controlled. Then the use of DNA would not be unidirectional read only thing modifiable only by random errors but would be more like library where the librarians and users can update and modify the content. Would it not be more efficient and faster? Why would the unidirectionality be preserved over billions life just to satisfy the DNA paradigm of the second half of XX century?
Another little example - the bacterias can adopt faster to the hostile environment (like antibiotics) by SHARING the genetic information even between the species.
The only cell we inherit is the egg cell from our mother. What process is there for this *memory*?
Egg cell is very alive and HUGE compared to the DNA. It is a hole through which a proverbial track can be driven through. And you have forgotten that spermatozoid also is a living cell with living and very active protoplasm.
You see, DNA is never really in charge and never on its one - it is always handled in a organized manner by the surrounding living cells. If they can read what they want, very likely they have capacity for occasional write in or switch on/off.