-
Notifications
You must be signed in to change notification settings - Fork 1
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Do we include NULL contributors in AID seed? #21
Comments
After thinking about it a bit more, I think they should be included in the seed (but they should be ignored during aggregation).
But the seed is a property of the bucket, so any encountered AIDs have to contribute to it. If we find a way to compute the seed while we digest data, then we can drop |
On the assumption that it's the value contributed by the AID that's
|
Why should any encountered AID contribute to the seed, if that AID doesn't otherwise contribute to the bucket? |
Hmm, let's consider this scenario: There are 100 rows of shape |
As much as I would like to use this opportunity to get rid of some code, I still think it should affect the seed. |
It passes the low count filter (and all AIDs contribute to the seed for the low count filter), but we don't produce an aggregate, because we have insufficient data for the aggregate. |
The main question is do we need to have the same seed for all aggregators (including LCF) or not. |
There are a number of different things being discussed here, so I'm a bit confused. The set of questions seem to be:
Regarding 1, what would cause an AID to be NULL? Regarding 2, can we avoid this question by always knowing what the actual AID is? Regarding 3, this question doesn't arise for Publish AFAIK, and I think it is premature to ask it for the other variants. In fact, none of this really matters for Publish... |
The question is not for NULL AID but for NULL contribution coming from a (non-null) AID. In the example below, which AIDs will be used for the AID noise seed for
|
Ah, I thought you meant the AID itself was NULL, not the aggregate. My intuition is that the AIDs with NULL contribution to the aggregator should be included in the seed as well as the LCF computation. And I can't think of an attack that would exploit this. And I presume it is simpler to just include the AID in all cases (no special cases to deal with NULL). So let's go with including all AIDs regardless of contribution to the aggregator. |
Actually, it is easier to exclude NULL contributions (simpler to ignore stuff sooner, rather than later). |
Closing this as NULL contributions are already included in the seed. |
If an AID contributes with
NULL
, do we include that AID in the seed material?The text was updated successfully, but these errors were encountered: