Loading…
Loading…
About the Project
Problem Genome Project was created by Beth Altringer Eagle and originally funded by the Harvard Initiative for Learning and Teaching (HILT) in 2013–2014. It was designed to collect and make accessible basic information from top experts on important problems, such as the Grand Challenges in engineering or the Millennium Development Goals. The idea came from seeing aspiring entrepreneurs (often but not exclusively students) repeat the same entry-level research on the problem areas that interested them. Interviewing experts one at a time, however, for this knowledge was an idea ahead of its time.
The catalyst was a student team in Prof Altringer Eagle’s class, working on a safer, cheaper syringe, whose half-hour call with a WHO expert ended with him describing the bookshelf in his office, littered with most of the designs they were inventing. The syringe is already very cheap and supported by a global system that will not easily transition to a new solution. You need a better problem, he said, and outlined one; they pivoted to a much better project on the vaccine cold chain. PGP is about making the equivalent of that call accessible to more people embarking on a significant project, so they can build atop what came before and get on with the more interesting questions much faster.
Fast-forward to 2025 at Dartmouth, and new technologies and support from the Altringer Eagle Lab in the Thayer School of Engineering at Dartmouth College make the original idea possible.
Problems are often tackled within fields whose experts seldom talk to one another. Problems across fields can share commonalities. A sensor failing in a Brazilian pasture and one failing on an ocean buoy can have a common cause. A similar mismatch between rules and technology holds back digital therapeutics and traditional medicine. A researcher inside a health system and an agency outside it can describe one crisis in different terms.
The Problem Genome is a living collection of 613 unsolved problems across 20 domains, each carrying genes along eight dimensions, about 80 genes in all. The genes show which problems are stuck in similar ways, across fields one expert would rarely see together. Since 2013, it has evolved into a human–AI collaboration project at Dartmouth, human ideated and designed, built for researchers and project-based learners: our system tags at scale while people steer, iterate, verify, teach, and translate.
Method
The Problem Genome system tags each problem with structured attributes called genes.
The collection begins with expert sources. The Problem Genome system reads research agendas, standards roadmaps, agency gap reports, review papers, and workshop proceedings, and works to identify unsolved problems with their constraints and context and describe them along the dimensions that we use to translate them, including things like the domain, binding constraints, failure mode, scale, what it will take to solve the problem, direction over time, and where someone might start if they wanted to work on it for a project this term.
Write-ups are checked against sources before they enter the collection and read again periodically in full over time; updates and corrections get date stamped in the problem entry.
Sources are graded by how close they are to the experts. Tier 1 is a primary expert source, such as an agency’s own report or the paper itself; Tier 2 and 3 are analyst, conference, or community sources; a write-up marked “needs sourcing” is waiting for a better one. The set of genes is reviewed against the data at milestones.
Once problems carry genes, the ones that share them across unrelated fields can be found. A sensor that works in the lab and fails in the field appears in 92 problems across 17 fields, from ocean monitoring to medical diagnostics to construction, and a fix from one of those fields may carry over to another.
Who describes a problem also shapes it. Perspectives compares 18 pairs of write-ups, one from inside the system and one from outside. It is an exploratory analysis.
Classification
Every problem is described by genes in eight categories. Choose any gene to see the problems that carry it.
Every card is built from the same six parts. The painted swatch is the problem’s field. When the title names its subject, that one word takes the field’s color. The ink marks around the title belong to its family and show why problems in that family get stuck. The dot at top right shows how solid the sourcing is. The plain words at bottom left say why the problem is stuck. The pale ground shows the kind of hard, with a band for each binding constraint and a tick for each unusual one. Once you can read one card you can read all 613.
learn the whole notation →Before these cards, the collection drew each problem as a round glyph. See the round glyphs of early 2026 →
License
The collection is licensed CC BY-SA 4.0. Use it, teach with it, build on it, and sell what you build. There are two conditions. Credit the collection, and license whatever you build from it the same way, so it stays open for the next person.
Download all 613 problems as JSON → The file holds every write-up with its genes, source, and text.
Cite it as: Problem Genome Project, Altringer Eagle Lab, Dartmouth College, www.problemgenome.com, CC BY-SA 4.0, accessed [your date]. The license covers our write-ups. It does not cover the sources they cite, which keep their own terms. Check those before reproducing anything from them.
For example: