What does a lonely record look like?
Imagine adding a card to Culture Atlas for a tradition. The card has a name and description, but it does not connect to a place, community, language, person, story or useful source.
Technically, the record exists. But it is not doing what I want Culture Atlas to do.
I started calling these lonely records. In graph language, people sometimes call a record with no connections an orphan. A record with only a few useful connections can be called sparse.
The Atlas can point to what is missing
This gave me an idea I had not expected when I started.
Culture Atlas does not only have to answer questions. It can help create research questions.
If a tradition has no source, that becomes a question: where can I verify it? If a person has no connection to the movement they are known for, that becomes a research task. If a story is connected to a state but not to the community or language that carries it, maybe the record needs more work.
A gap is not automatically a mistake
I have to be careful here.
Just because the database does not contain a connection does not mean the connection exists in real life. The worst solution would be to fill empty spaces just because I want the graph to look complete.
The gap is a reason to investigate, not permission to guess.
So I need a research queue
The project has a research queue for graph gaps. I think of it like a to-do list made by looking at the shape of the Atlas.
Instead of saying “research Indian culture,” which is impossibly large, the system can point to smaller questions: this record is isolated; this one has weak source coverage; this connection needs checking.
That turns a huge project into questions I can actually work through.
Connections need evidence too
I used to think mainly about sourcing facts inside a record. Now I am also asking whether the line between two records needs evidence.
If I say a tradition is practised by a community, that relationship is itself a claim. Before adding a new connection, the project rules say to check both records, the type of relationship, confidence and evidence.
That sounds technical, but the basic question is simple: How do I know these two things belong together?
A lonely record is not automatically a bad record
Some subjects may genuinely have fewer documented connections than others. If I treat every sparse record as a problem, I could start inventing links simply to make the graph look healthy.
The research queue should create questions, not answers. A gap tells me where to investigate; it does not tell me what I am supposed to find.
Absence needs context
History has made me more careful about missing evidence. If I cannot find a connection, that might mean the connection does not exist—but it might also mean it was never recorded, the source is inaccessible, or I have not looked in the right place yet. Geography can create similar blind spots when documentation is concentrated in some regions more than others. So an empty part of the graph should start an investigation, not finish one.
A gap is a question, not an answer
An incomplete graph is not a failure if I can see where it is incomplete.
I would rather have a visible research gap than add a confident-looking connection I cannot support.
What I'm figuring out next
Finding a gap is only the start. Next I need a repeatable way to take one real record, research it, compare sources and decide what is strong enough to add.