Methodology

The Sphere is a grid: 12 fields of knowledge along the circle, 8 depth levels from the centre outward. That is 96 cells. Every skill you record lands in exactly one cell, and the hours behind that skill fill it.

The 12 fields are: Language & Communication, Mathematics & Statistics, Natural Sciences, Digital & ICT Skills, Engineering & Manufacturing, Health & Safety, Business & Administration, Law & Public Administration, Social & Cultural Skills, Arts & Humanities, Transversal Skills, Green & Sustainability.

The fields are adapted from the ESCO knowledge pillar. They are not ESCO itself: we merged and renamed branches so that a reader can place a skill without consulting a classification manual.

The 8 depth levels are the levels of ISCED 2011, the International Standard Classification of Education kept by UNESCO. The scale is international by design, so a level means the same thing whichever country the knowledge was gained in. Each level is shown with the US degree anchor used in the interface:

• 1ISCED 1 — Primary educationSchool basics
• 2ISCED 2 — Lower secondarySchool basics
• 3ISCED 3 — Upper secondaryHigh school
• 4ISCED 4 — Post-secondary non-tertiaryCertificate
• 5ISCED 5 — Short-cycle tertiaryAssociate
• 6ISCED 6 — Bachelor'sBachelor's
• 7ISCED 7 — Master'sMaster's
• 8ISCED 8 — DoctoralDoctoral / Frontier

A skill learned on a course takes the level of the programme it belongs to, as UNESCO maps that country’s programmes onto ISCED. A skill learned at work has no programme, so its level is read with the descriptors of the European Qualifications Framework: how deep the knowledge is, how complex the work is, and how much responsibility and autonomy it carries — for example, whether you lead a team. EQF is the rubric; the level itself is ISCED. Today the level is judged from the skill itself; how you use it at work is not yet part of that judgement.

Level numbers stay in the data model and the API. In the interface we show the degree anchor instead, because it is the register a US reader already thinks in.

Each depth level has a capacity in hours — the amount of deliberate time it takes to hold knowledge at that depth. The capacities are design constants, chosen to be comparable to the O*NET SVP preparation scale:

• School basics50 h per cell
• School basics150 h per cell
• High school400 h per cell
• Certificate800 h per cell
• Associate1,500 h per cell
• Bachelor's3,000 h per cell
• Master's6,000 h per cell
• Doctoral / Frontier10,000 h per cell

Across all 96 cells that totals 262,800 hours. That number is the denominator of the percentage you see.

To state it plainly: "All Human Knowledge" is the frame of the map, not a measurement of humanity. Nobody has counted what humanity knows, and this grid does not attempt to. The percentage is the coverage of your own map against that frame — which is why the word is mapped, not mastered.

• Activity logs — the duration you record, minute for minute.
• Books — pages divided by 30 pages per hour.
• CV experience and education — 167 hours per month of the period, distributed across the skills attached to that entry, weighted by how central each skill is to it.

Hours are facts about your record, so they are always shown raw and uncapped, even where the percentage stops counting them.

• Each cell has a cap. Hours beyond a cell's capacity still appear in your hour totals, but they stop moving the percentage. Depth in one place does not read as breadth.
• Placement of a skill into a field and a depth level is done by a language model. It is consistent across users, but it is a judgement, not a measurement.
• Nothing here is verified by a third party yet. The record is yours, and it is self-reported.

Pareto skills answer a narrower question than the Sphere: of everything you have recorded, which few skills carry most of your earning weight.

• Each skill is matched against occupations ranked by salary. The three best-paying matches are blended, each weighted by how strongly the skill matches that occupation.
• The blend is then seniority-adjusted, using the hours behind the skill as the proxy for how far along its curve you are.
• Every skill carries its share of that blended earning weight. Shares are normalised across your record, so they always sum to 100%.
• Sorted by share, the Pareto list is the smallest set of skills whose shares reach 80% cumulative weight. Everything below that line is the long tail.

The list is about concentration, not worth. A skill outside it is not weak; it simply carries less of the earning weight in your current record. Salary figures are market ranges, not offers, and the matching is a judgement made by the same classification model used elsewhere.

The graph behind the Graph page has one kind of node: a skill. The 12 fields of the Sphere are its roots. Between a root and your skills sit the groups of the ESCO skills hierarchy, about 650 groups in four pillars. Every node carries where it came from: ESCO, your own record, or a classification model.

• A parent is always an existing node. A skill of yours is placed under a group that already exists; when no group fits with confidence, the skill stays directly under its field. Nothing is invented to hold it.
• A skill can belong to more than one group. One edge is primary: it decides the field, the path to the root, the position on the page and where the hours are counted. Further parents are drawn dashed as “also part of” and carry a share of the skill, not a copy of its hours.
• Two spellings of one skill stay two nodes. The graph does not merge your record; reading them as one skill is a separate layer, still to come.
• Prerequisites are a different relation. They exist in the graph and on a skill’s card, but the personal graph shows containment only.

The graph is derived data. It is rebuilt from the ESCO hierarchy, the shared taxonomy and the placement decisions, and a rebuild never changes your hours or your skills.

Methodology v2.2 (2026-09). Version 2 introduced the per-cell cap, a shared classification taxonomy across all users, and the shift in wording from mastered to mapped. Version 2.1 rebuilt the skill graph on the ESCO hierarchy with one primary parent per node. Version 2.2 states ISCED as the scale of depth levels and EQF as the rubric for skills learned at work. When the capacities, the grid, the graph model or the classification model change, the version changes with them.