See 0014166 Feature Request
I think I also post here for discussion on possible solution strategies etc.
I am looking for a systematic way to handle uncertain parent-child relationships in cases where historical sources only allow probabilistic estimates.
Typical situation:
A man was married three times. Based on chronology (estimated birth year of a child, marriage dates, wills, etc.), it is unclear which wife is the biological mother. The child could most likely belong to the second or third wife, but the first wife cannot be completely ruled out either. I would like to be able to assign probability values to each possible mother — for example: 60 % second wife, 30 % third wife, 10 % first wife.
In ancestor charts, ancestor lists or reports, all possible mothers (or fathers) should then be displayed together with their respective probability percentages. These probabilities should also propagate to the ancestors of these possible parents. In cases of further uncertain filiations in the line or in implex/pedigree-collapse situations (where the same person appears as ancestor through multiple lines), the probabilities should be automatically recalculated (e.g. multiplied or adjusted using Bayesian methods).
In my own research, such uncertain cases accumulate across many generations in certain ancestral lines. Due to a lack of additional sources, these uncertainties cannot be substantially reduced. This makes systematic listing and evaluation of even parts of the family tree increasingly difficult and unclear.
Current state AFAIK:
The GEDCOM standard (5.5.x and 7.x) unfortunately does not provide any field for filiation probabilities. Only qualitative indicators such as PEDI, FAMC-STAT (PROVEN/CHALLENGED/DISPROVEN), QUAY or notes are available — quantitative percentages and automatic propagation are not supported.
Gramps is known for being highly extensible through custom attributes, notes, reports and addons. In theory, this feature could be implemented as an addon or extension. However, I personally lack the programming experience to create such an addon myself.
I would like to propose this as a possible future feature or addon idea. Has anyone else encountered this need? Are there any existing approaches, partial solutions, or thoughts on how this could be implemented cleanly in Gramps?
2 Likes
Perhaps take a look at the links in this text from the Genealogy Discord server (2026-03-05):
Hey everyone đź‘‹ I'm Noah, a developer and family archivist in Des Moines. I've been building **GLX (GENEALOGIX)**, an open source YAML-based spec for storing genealogical (+FAN and beyond) data in git repositories.
GLX treats your family archive like source code: versioned, portable, fully yours. It's designed to do what GEDCOM can't: complete evidence chains, inclusive relationship types, and oral histories as first-class data alongside official records. It's still in beta and very much a work in progress, and that's intentional. Community feedback is what's going to determine where the spec goes from here.
I'm also building **Oracynth**, a web app on top of GLX, with early access opening soon — but the bigger goal is GLX becoming an open standard any tool can build on.
đź”— GLX spec: https://genealogix.io
đź”— Oracynth: https://oracynthsoftware.com
Happy to answer questions here, or:
- 🗓️ Find me at **RootsTech this weekend**
- đź“ž DM me to chat or set up a quick call
What pain points have you hit with GEDCOM or existing formats?
See also that project blog page about its YAML assertions/evidences/conclusions format: Introducing GENEALOGIX: An Open Format for the Family Archives We Actually Have
The YAML files mentioned in this project’s documentation could probably be replaced by notes (YAML notes?) and attributes in Gramps.
This is a classic graph problem rather than a tree problem, and it would be far easier to solve if Gramps could store roles, relationship types, and filiations as triplets — or more realistically as quadruplets:
{Type, Probability, Date, Source}
This idea has come up several times in the past when discussing other relationship‑modeling limitations. The core issue is that genealogical data is not inherently tree‑shaped; it is a probabilistic, multi‑edge graph with uncertainty, implex, and multiple competing hypotheses.
If relationships were represented as explicit graph edges with properties (including probability), then the probability between two entities could be treated as a mathematical operator usable in analysis — including propagation, Bayesian updates, and handling of pedigree collapse.
Personally, I handle this kind of modeling quite easily in Obsidian, using YAML‑based edges with probabilities and dates. The domain is different, but the principle is identical: relationships are edges in a graph, and probability is just an attribute of the edge. This makes it straightforward to run analyses that are currently difficult or impossible within the constraints of GEDCOM and the present Gramps data model.