Does PUFFIN prevent some transactions from taking place based on what happened on another site?
Yes, the PUFFIN is based on previous ad transactions that may have happened on any site.
However, a bidding function is never made aware of whether it lost an auction due to PUFFIN or due to being outbid on the current site.
Is the PUFFIN available to parties in the client-side auction?
No. Making PUFFIN available here could encourage sites to add too many marginal people to the interest group and then rely on PUFFIN to only bid on some of them.
Is PUFFIN a privacy protection measure?
A significant risk for browser-enabled advertising is that deceptive and fraudulent sites will be able to monetize ad impressions from desirable users.
Each high-value user will be carrying around a bunch of high-revenue ads in their browser, and all that a deceptive site has to do is somehow get one of those ads to bid.
In-browser ads give bad actors a powerful incentive to send out a large number of links to their sites in hopes of getting some high-value traffic. This could take the form of:
-
harmless but annoying tactics like chumboxes
-
ToS-violating but legal techniques like email and forum spam
-
Malware
Without revenue protection for high-engagement sites, TURTLEDOVE/FLEDGE help support malware by making it easier to monetize.
When is the PUFFIN applied?
In the FLEDGE origin trial, the contextual clearing price can be used as a floor for the FLEDGE auction, and the contextual ad can be rendered in a regular iframe if the FLEDGE auction returns no...results that beat the floor. The PUFFIN would be applied as an additional floor at this stage: the interest group ad would need to beat both the current winning contextual ad and the PUFFIN.