Index Counts in Federal Criminal Cases

[zvisrosen]

As of now, the system does not seem to track criminal counts or provide a way to search for them. This would be a useful feature for finding cases, keeping track of what actions are being brought in what courts, having alerts for particular prosecutions, and for all sorts of other research.

[johnhawkinson]

To be a little more particular, criminal counts are tracked in the Parties table.

Let's take https://www.courtlistener.com/docket/65760207/parties/united-states-v-napolsky/ for an example.

If we view its parties in the API at https://www.courtlistener.com/api/rest/v4/parties/?docket=65760207, we have a party_types object with a criminal_counts array;

"criminal_counts": [
                        {
                            "id": 6742150,
                            "name": "17:506(a)(1)(A) and 2319(b)(1); 18; Section 2 and 3551; et seq- Criminal Copyright(1)",
                            "disposition": "",
                            "status": 1,
                            "party_type": 13229045
                        },
                        {
                            "id": 6742151,
                            "name": "Title 17, United States Code, Section 506(a)(1)(A); Title 18, United States Code, Section 2319(b)(1),2 and 3551 et seq.-CRIMINAL COPYRIGHT(1s)",
                            "disposition": "",
                            "status": 1,
                            "party_type": 13229045
                        },

but unfortunately we cannot search it.

@albertisfu suggests that we could index it to permit queries like

https://www.courtlistener.com/api/rest/v4/parties/?party_types__criminal_counts__name__contains=Copyright&filter_nested_results=true

to filter nested fields, similar to #4729.

[mlissner]

Thanks. Let's keep this issue here, and if anybody else is looking for this, they can chime in. I don't think we'll open up this field for queries via API at the moment due to performance, but if somebody has a few queries they want done on the back end, we can probably do that.

[lukerosiak]

This would be useful to me and, I would assume, others. People often want to examine cases of a particular type (homicide is more interesting than assault) or follow the application of a particular law, such as RICO.

Performance aside, the charges also don't seem to be standardized. The below example has two different formats in one criminal case, while the example above is a third:

['name': '15:78j(b), 78ff; 17 C.F.R. 240.10b-5 Securities Fraud(1-2)',
       'disposition': '',
       'status': 1,
       'party_type': 16360518},
      {'id': 6959077,
       'name': '18:1343 Wire Fraud(1s)',
       'disposition': '',
       'status': 1,
       'party_type': 16360518}]

In PACER, one can go to Criminal Cases Report and there is a dropdown for Citation which appears to list all possible criminal offenses (e.g. "18:13-7420.P Racketeering, Bribery") and which can be used to filter cases. I don't know if this field is reliable, available and makes its way to CourtListener.

[mlissner]

Interesting. We haven't done anything to normalize this field, but if we made it searchable, we probably would need to. The challenge with this issue is that we have:

One case with...
   Multiple parties, with...
       Multiple offenses.

So if you compress all that down into a search engine, you wind up with something like:

Case Name	Case ID	Parties	Offenses
Super v. Villain	1	Superman, Robin, Bad Guy	Homocide, Assault, etc

Which isn't great, because it separates the parties from their offenses, but it would make them searchable and alertable. You just wouldn't know who did what, since that would be mushed together.

Maybe it's not so different from attorneys and firms, which we do make searchable:

One case with...
    Multiple parties, each that has...
        Multiple firms, where there are...
             Multiple attorneys.

So I guess that's all to say that it's doable, just not easy, and it'll come with compromises.

[lukerosiak]

I don't think the fuzzy relationship between parties and cases is a problem. In addition to the analogous issue with attorneys, the "doppelgänger issue" already makes it all a bit fuzzy, but ultimately most people just want to locate a case involving a certain type of conduct. They will have to read it if they want to be sure about which defendant is which, but multi-defendant cases where one party faces only minor charges are an edge case anyway.

Do you know where that "Citation" drop-down in the Criminal Case Report view of PACER maps to? That would avoid the need for standardization. There is a trade-off there, which is that only one item can be selected, but I believe it would be used for the top-charged offense, and that is what most people would be interested in anyway, so I think having one pre-standardized field would be better than a messy one-to-many.

[mlissner]

I don't think the fuzzy relationship between parties and cases is a problem.

I agree. I think that should be fine.

Do you know where that "Citation" drop-down in the Criminal Case Report view of PACER maps to?

No idea. Judging by the messiness of our data, it's not the thing being used on the docket sheet. Sometimes there's data that's internal to PACER that we can't get out except, perhaps, through the FJC IDB.