Climbing is what makes a hike hard, and flat miles hide it. So every 500 feet of climbing counts as one extra mile.
A flat 6-mile walk is 6.0 effort miles. Little Si — 3.7 miles, 1,300 ft up — is 6.3. Your legs already knew those two are about the same amount of work.
Prefer kilometres? It's the same number, converted. Multiply effort miles by 1.609 to get effort km. That works out to about one extra km for every 95 m of climbing. So the flat walk (9.7 km) is 9.7 effort km, and Little Si (6.0 km, 395 m up) is 10.1. Choosing km changes how your plan is written, never what it asks of you.
That's the formula: distance + feet of climbing ÷ 500. It isn't tuned or personalised. It's the same for every route and every person.
One addition: a pack. On a day you carry a loaded pack, the effort counts for more: 10% more with a light pack, 20% more with a heavy one. If your goal is an overnight trip, the hike itself is counted with a heavy pack.
This is what lets you compare very different walks. Say you spend your lunch break going up and down a staircase. You only cover half a mile (0.8 km), but you climb 1,130 feet (345 m). That's 2.8 effort miles, not 0.5 (4.4 effort km, not 0.8). It's how someone with no car and no trailhead nearby can train for a mountain on city sidewalks. And it's why the app measures progress in effort miles, not distance.
Training plans hurt people when they ramp up too fast. So these limits never bend, whatever date you give the app.
The plan also aims for you to finish one day that's 80% of your goal hike, at least two weeks before the hike itself. That one is a goal, not a limit. When there isn't time to reach it safely, the limits above still win.
When you start a plan, the app works out the earliest date you could be ready and fills it in for you. If you pick an earlier date, it tells you so. Either the plan comes up short, and it shows you by how much, or the gap is too big and it won't build the plan at all.
What it won't do is squeeze the training into fewer weeks to make your date work. Squeezing is how training plans cause injuries, and every rule above is there to stop it. If the honest answer is “you need more weeks”, that's the right answer, not a bug.
The app uses an AI language model for writing: explaining what a week is for, and describing the routes it suggests. It never uses AI to decide anything about your training.
Every number you see comes from the rules above, worked out by plain, testable code. The AI writes afterwards, and it may only repeat numbers your plan already has.
That's checked, not just hoped for. Before you see anything the AI wrote, every number in it is checked against your plan. If one doesn't match, the whole paragraph is thrown out. An AI that's confidently wrong about a distance could get you hurt, so it never gets asked.
Effort miles count distance and climbing. They don't count going downhill, which is often what wrecks people’s legs on a long hike. So a hike with a big descent can score lower than it should. When a route is like that, the plan tells you in words instead of pretending the number covers it.
The 500-foot rule (about 95 m per km) is a rule of thumb, not a law of nature. It's a good average for the kinds of terrain this app covers. It's the same for everyone on purpose: a number tuned to each person would be one nobody could check.
Spot something that looks wrong? Please tell us at kate@thepondsedge.com.
Training carries risk covers what this plan asks of your body, and what the app can't see.