Christian Lassen Dam

Three things overlap here, and none of them on its own would be enough.

The engineering. I am doing an MSc in Mechanical Engineering at Aalborg University. Everything a cycling calculator does is a force balance: gravity pulling a rider and bike down a slope, rolling resistance at the contact patch, aerodynamic drag rising with the square of speed, and the drivetrain losing a few percent on the way through. Resolving forces, fitting a curve to noisy measurements, checking whether a model still behaves at the edges of its input range - that is coursework I get graded on, not something I picked up from a forum thread. It is also why I am comfortable saying when a model breaks, which is the part most calculators skip. I wrote the long version of this in the mechanical engineering of Zwift performance.

The riding volume. I have been on Zwift for years and I ride the Tron bike. You cannot buy that bike. You unlock it by finishing Zwift's Everest challenge, which means climbing 50,000 vertical metres in game. Alpe du Zwift gains 1,036 m per ascent, so that is the equivalent of a little over 48 laps of the Alpe. I mention it because it is the only honest proof of volume I have: it does not make me fast, it means I have spent a very large number of hours inside the thing I am writing about. I know what a Zwift race actually feels like in the first two minutes, how the draft behaves in a big group, and why your in-game weight matters more than almost anything else you fiddle with.

The structured training. I am training for a full-distance Ironman, with Ironman Copenhagen as the goal race. That is where the training side of the site comes from: building a week around two hard sessions instead of seven medium ones, protecting the long ride, fuelling a four-hour indoor block, and fitting all of it around a degree. Fitting it around a degree is its own problem, and I wrote about that in having time to Zwift during university.

I use language models when I write. It would be dishonest to run a site like this today and pretend otherwise, so here is exactly what that means.

Where I use them. Drafting and structuring long articles, turning a messy set of notes into an outline, tightening a paragraph that has gone baggy, catching the third time I have made the same point, and doing the boring parts of building pages. On a 2,000-word article that is genuinely useful, and it is why this site has more written explanation behind its numbers than it otherwise would.

Where I do not. Not for the numbers. The equations, the regression fits, the climb data, the worked examples and every conclusion drawn from them are mine, and each one is checked by a human - me - before it is published. A model does not decide that a quadratic is the right shape for the Alpe data, does not get to state that the climb gains 1,036 m, and does not get to tell you what a result means for your training. I also will not let a model hand me a citation. If I cannot find the source myself and confirm it is real and live, the claim either comes out or gets rewritten as my own reasoning, clearly labelled as such.

Where the judgement sits. With me, at four specific points: what gets published at all, which numbers go in, which caveats travel with them, and when something already published turns out to be wrong. Nothing goes live that I could not defend if a reader emailed me and asked where it came from. Google's guidance on this is reasonable and I agree with it - the problem was never that a tool helped with the writing, it is unchecked, unedited output shipped at volume with nobody accountable for it. My name and my email are on every page precisely so that there is somebody accountable.

The two climb calculators are the core of the site. The Alpe du Zwift calculator covers the 12.2 km, 1,036 m, 21-hairpin climb that most riders use as their benchmark. The Ven-Top calculator covers Zwift's version of Mont Ventoux in the France world, roughly 20.9 km and about 1,534 m of gain, which is a different problem entirely. At the same W/kg the two models here put Ven-Top roughly 35 to 40% slower than the Alpe, and once you are over the hour mark pacing and fuelling start to matter more than raw W/kg. If you are choosing between them, the Alpe versus Ven-Top comparison puts them side by side.

Around those sit the supporting tools - FTP estimation and racing category - plus the Zwift climbs reference, which is exactly what it sounds like: length, elevation and average gradient for the climbs people actually ride, in one table, so you do not have to go looking. I would rather that page be short and correct than long and half-verified, so it grows slowly.

The rest of my time goes into the articles, which fall into four buckets: training, racing, setup and data analysis. The data analysis pieces are the ones I most want people to read, because they show the working behind the calculators rather than just the output. More on why the site exists at all is on the about page.