I started writing a reply, but @Nagalim beat me to it. I still want to post the reply. It might help.
Maybe @Nagalim would be better suited to explain this, because he has in-depth knowledge from not only designing it, but implementing it as well.
On the other hand I might be suited well because misunderstandings on my side can help others to develop an own understanding.
So back to my understanding:
the unseeded auction is in a test run situation. That’s why using small amounts is recommended. The software created by @Nagalim is suited to handle the auctions automatically. Participants who send NSR and NBT to the auction addresses have their input factored on the NSR and NBT side of the auction.
At the end of the auction the quotient of the total auctioned amount of NBT and total auctioned amount of NSR determines the price point NBT/NSR.
Participants can trade NSR for NBT or vice versa with other participants.
Participants whose bids (ratio of NBT/NSR) are below the price point, will have less NSR and more NBT after the auction compared to what they placed in the auction. They traded NSR for NBT, because the “market” considered NSR more valuable than the participants being below the price point did.
Participants whose bids (ratio of NBT/NSR) are above the price point, will have less NBT and more NSR after the auction compared to what they placed in the auction. They traded NBT for NSR, because the “market” considered NSR less valuable than the participants being below the price point did.
As soon as this type of auction is held as seeded auction (Nu being a participant), Nu can convert NBT to NSR or NSR to NBT - depending on the liquidity/market situation of Nu.
This could be groundbreaking if it would be held continuously with sufficient participation.
The motivation for non-Nu participants can clearly be arbitrage between the auction and exchanges.
The motivation for Nu is to have an NBT/NSR exchange gateway beyond exchanges