DigitalEPO is written by Hayden Rees and covers endurance-sport data: what tracking platforms measure, where those measurements fail, and what can honestly be concluded from a recorded performance. The material is split between written reference pages and instruments that do arithmetic in the browser. Both halves answer the same question from opposite ends: what can a recorded number actually support?
The approach
Every instrument on the site prints its assumptions next to its output. A number produced by a hidden model settles nothing, because it cannot be examined or disputed. Showing the coefficients, the duration curve and the terrain conditions makes each result something a reader can argue with. An arguable number is more useful than a confident one.
The written pages follow the same rule. Where a figure is approximate, it is described as approximate. Where a mechanism has a known failure mode, the failure mode is stated rather than skipped. Endurance data is full of quantities that look precise and are not. A figure printed to two decimal places is not thereby accurate to two decimal places.
Reading a result, and stopping there
The instruments report what figures imply and where that sits against recorded human performance. Several ordinary explanations produce an impossible number, and arithmetic performed from outside cannot separate them, so the reading is where each tool ends. Naming a cause would require evidence that no outside calculation has access to, so none is offered.
Files are read, never written. An activity opened in the audit is parsed in the browser and never transmitted, and the reconstruction tools take typed figures instead of uploads. The archive describes what this address carried between 2013 and 2016, and describes that material instead of reproducing any of it.
Who writes this
Hayden Rees writes every page and builds every instrument here. The name is a byline, not a legal identity, and nothing on this site depends on who holds it: each result is reproducible from the figures and coefficients printed beside it, which is a stronger warrant than a biography would be. Corrections are welcome from anyone.
The working method is narrow and consistent. A claim about a platform is checked against that platform's own documentation and against files it has actually produced. A physical claim is derived from the model printed on the page, with the assumptions named. Where the two disagree, the disagreement is described instead of resolved by assertion.
What that covers in practice is cycling power reconstruction, GPS and barometric recording error, segment and leaderboard verification, and the export formats these platforms use. Anything outside that range is either left alone or linked to somebody who works in it, because the useful thing a small site can offer is a narrow subject treated properly.
Corrections
Errors in the physical model, the reference ranges or the platform descriptions are worth reporting and will be fixed. The contact page is the route. Corrections that change a published figure are made to the page itself rather than noted and left. A wrong number left standing with a footnote is still a wrong number.
The instruments are at tools, the platform reference at Strava data, and the domain’s earlier life in the archive.