This card has a fairly narrow use, but man can it be powerful in the right deck. This card's most obviously useful in decks like Industrial Genomics: Growing Solutions and Gagarin Deep Space: Expanding the Horizon where recursion of assets and archive control is useful.

But this card has some other powerful side effects as well. It's not just getting back your cards, it's that you're able to shuffle your R&D each turn. This can make cards like Medium, Indexing and other R&D lock mechanisms much less effective. Additionally, because this is neither unique nor expensive, it's not impossible to get a situation going where you are putting three cards back into your R&D every turn, reducing Agenda density substantially.

I can easily see this being the first in a series of cards that encourages and powers larger sized decks. Fun stuff!

Well, it makes Medium much MORE effective, since every turn runner can check X new cards from top of the R&D server. —
#konradh I'm not sure about that, actually. I'm no math major and I could be wrong, but if you're digging into R&D, you steal agendas and leave the non agendas (for the most part). That means the more runs you make, the deeper you go, the moren likely you are to see an agenda. If you shuffle each turn, you are re-setting those odds. It doesn't increase the number of cards seen but it resets the agenda density. I may be totally off on this though. Maybe a good question for someone with a math degree :-) —
It depends on what your goal is. If your concern is how easily they'll access agendas and they're accessing more than just the top card, then shuffling is bad for you because it refreshes ALL of the cards they're getting to see. If you're trying to avoid them knowing what you draw, then shuffling is good for surprises. The slow lowering of agenda density doesn't really do much. —
Assume that the runner has a medium on 3 and they run and show 3 ice. A shuffle will trade off 3 known to not be agenda cards for 3 possibly agenda cards. Adding cards back in that are non-agenda adds some tiny % likelihood that a possible-agenda is less likely to be an actual-agenda. Trading known-non-agendas for possibly-agendas adds a huge % to the likelihood the runner gets a successful score, roughly 20% per card. It's orders of magnitude better for the runner in the scenario I described. —