Yukon Solitaire: a short look-ahead is not enough
Yukon deals every card onto the table — there is no stock — and lets you lift any face-up card together with everything piled on top of it, in order or not. That one freedom makes Yukon a planning game through and through, and it gives this site's question a sharp answer: a player who looks only a little way ahead wins rarely, while most deals can be won by someone who plans the whole line.
Our Yukon bot is not a pure rule-of-thumb player. It searches up to 300 positions ahead on each turn, using only the cards it can see, and never undoes. It won just 4.5% of 10,000 deals (95% range 4.1% to 4.9%). A deeper planning search, allowed to know the face-down cards, proved 67.0% of 500 deals winnable (95% range 62.8% to 71.0%). A gap of more than 60 points, even with some look-ahead built in. We have no plain greedy bot for Yukon, so the table compares short look-ahead with deep search.
Yukon at a glance
| What we measured | Result | n | 95% range |
|---|---|---|---|
| Short look-ahead bot (visible cards, 300 positions per turn, no Undo): won | 4.5% | 10,000 deals | 4.1–4.9% |
| Deep planning search (knows hidden cards): proven winnable | 67.0% | 500 deals | 62.8–71.0% |
| Deep search: undecided within budget | 29.6% | 500 deals | 25.8–33.7% |
| Deep search: proven impossible | 3.4% | 500 deals | 2.1–5.4% |
| Cards the bot got home, all games | 7.4 of 52 | 10,000 deals | 7.2–7.6 |
| Moves in the bot's wins | 96.9 | 445 wins | 96.5–97.4 |
So between 67.0% and 96.6% of deals can be won with perfect knowledge. The bot's 4.5% is a floor, not a human win rate; people with Undo and patience do much better. The published study by Blake and Gent does not cover Yukon, so these figures stand on their own.
Yukon rules on spidersolitaireonline.us
- Layout: seven columns. The first has one card; the others have six to eleven, with the last five of each face up.
- No stock: all 52 cards start in the tableau.
- Foundations: four, by suit, Ace to King.
- Building: one rank lower in the opposite color.
- Group moves: any face-up card can be moved with everything on top of it, sequence or not. Only the card you pick up has to fit.
- Empty columns: a King only, with whatever is on it.
What planning buys in Yukon
Plan the whole pile, not just the top card
When you move a group, everything on it comes too, including cards that may block you later. Before a group move, check what you are carrying and where it will end up. The bot's 300-position look-ahead was not enough to see these knock-on effects; your plan can be.
Aim every big move at a hidden card
The bot ended its games with only 7.4 cards home on average, usually stuck with cards still face down. A group move that exposes a hidden card is almost always worth more than one that tidies the face-up cards.
The deal is not the problem
On average 1.6 of the 4 Aces start face down. The bot won 4.4% of deals with two or more hidden Aces and 4.5% with one or none — no real difference. What separates wins from losses is the order of play.
Decide which King takes the space
An empty column accepts only a King, and the King brings its whole pile. Choose the King whose move uncovers the most, and do not empty a column before you know which King is going there.
Expect a long line
The bot's wins took about 97 moves. Winning Yukon means working through a long, indirect plan, often moving the same cards more than once.
Undo is your deeper search
The distance from 4.5% to at least 67.0% is what deeper planning is worth. You cannot see face-down cards, but unlimited Undo lets you try a big move, see what it reveals and back out.
Features here
Yukon is free in any modern browser on computer, tablet or phone, with nothing to download and no account. You get large cards and a zoom control, Hint, unlimited Undo, Restart, Auto Play, drag-and-drop or tap-to-move, and a tracker for wins, time and moves. Sound, fullscreen and table color are in the settings.
How we ran the study
The spidersolitaireonline.us team simulated Yukon in September 2026 on the rules this site's game code uses.
- Read the game code: seven columns with the last five cards face up, no stock, opposite-color building, group moves of any face-up card and what is on it, Kings only in empty columns.
- Wrote the game in Python. Yukon and Alaska share one program, since only the building rule differs.
- Dealt numbered hands from seeds: 1 to 10,000 for the bot and 1 to 500 for the deep search.
- Used two players. The bot looks up to 300 positions ahead each turn on visible cards, scoring cards home, hidden cards and good or bad links, and never undoes. The deep search is depth-first, knows every card and had up to 50,000 positions per deal.
- Checked for published figures. Blake and Gent (2026) do not report Yukon.
- Gave the uncertainty. Percentages use Wilson score intervals; averages use normal intervals.
- Stated the limits. The bot is weak at Yukon; the search sees hidden cards; nearly 30% of search deals stayed undecided.
The code uses only the Python standard library and ran across 19 processes: about two minutes for the bot's 10,000 games, about eight for the 500 search deals. Claude, an AI assistant made by Anthropic, helped write and review the code and draft this page. The figures come only from running the code.
Yukon questions
Why can I move cards that are out of order?
That is Yukon's defining rule: only the card you pick up must fit.
Is Yukon harder than Klondike?
For simple play, far harder: 4.5% here against 42.3% for our Klondike bot. With deep planning the gap shrinks, since at least two thirds of Yukon deals can be won.
Can every Yukon deal be won?
No. Our search proved 3.4% impossible, even seeing every card.
What should I try next?
Alaska keeps the group moves but builds in suit. If you like moving piles, two-suit Spider is next. All games are on the games page.
By Admin, spidersolitaireonline.us (Solitairen OOD). Research run September 2026.