An incomplete Skyscrapers check shows guidance only and never costs a mistake. The mistake counter is only affected when you press Check skyline on a fully filled 16-lot grid that still breaks a row permutation, column permutation, or shown edge clue. The first incorrect completed check adds one mistake and keeps the grid editable so you can repair the attempt, while the second incorrect completed check deadlocks the run until you press Restart. This two-tier policy is the single most misunderstood mechanic in compact logic games, because most Latin-square titles charge a penalty for any premature Check press. The published Skyscrapers Puzzle deliberately separates confirmation from correction: filling all 16 lots is the formal signal that you have committed to a grid, so anything earlier is exploration rather than a finished answer. The verified FAQ makes this explicit — an incomplete check only displays guidance, an incorrect completed grid can still be repaired, and a second incorrect completed grid freezes the run and requires Restart. The 1,000-point target only triggers on the single unique valid grid, so the check rules are designed to let a careful player try, repair, and finish without wasting a try on a mid-progress sanity check.
Does an Incomplete Check Cost a Mistake?
No, and the distinction is exact. A Skyscrapers board is considered "incomplete" until every one of the 16 lots has a height from 1 through 4. The moment any cell is empty, pressing Check skyline shows a small reminder that your grid is not finished. That reminder carries no mistake penalty, no scoring effect, and no terminal state. It exists so you can confirm row uniqueness, column uniqueness, or clue progress while you are still placing numbers. The verification methodology that built the fixture explicitly tests this harmless incomplete state, confirming that early checks are inert rather than destructive.
The phrase to remember is "completed check." A check becomes a completed check only after the 16th lot has a height assigned. Until then, every press is effectively a status query that never alters the score, the mistake count, or the run's liveness. Treating early Check skyline presses as free guidance rather than a gamble is the simplest way to keep the run alive while you are still reasoning about the rows and columns.
The Four Things a Check Can Return
Once you know what each possible check result does, the whole run becomes easier to plan. The board can return four distinct outcomes from a single Check skyline press, and only one of them finalizes your score.
| Board state at Check | Mistake change | What you see | Next state |
|---|---|---|---|
| Incomplete (any cell empty) | None | Reminder to fill all lots | Still editable, still no penalty |
| Complete and valid | None | Solution confirmation | Terminal: 1,000 points, input frozen |
| Complete with first error | +1 mistake | Highlighted conflict | Still editable for repair |
| Complete with second error | +1 mistake (total 2) | Deadlock notice | Terminal: only Restart can change the outcome |
The terminal states matter because they freeze the entire board. Once the correct grid is confirmed or two mistakes are reached, the height keys, the Delete and Backspace clear keys, the arrow-key cursor, and additional Check skyline presses all stop having any effect. Restart is the only action that clears a terminal state and restores the same empty 4 by 4 grid, the same five clues, a zero score, and a zero mistake counter so the next attempt begins from the exact same starting position.
How to Place Heights and Run a Check
The mechanics are intentionally simple because the difficulty lives in the logic, not in the controls. The same pointer, touch, and keyboard access apply on desktop and mobile, and a narrow viewport is preserved through zero-minimum fractional columns so the four-wide grid stays readable on a phone screen.
- Select a lot by tapping or clicking it, or by moving the cursor to it with the arrow keys.
- Enter a height from 1 through 4 by pressing the matching digit key or by tapping the on-screen number control for that cell.
- Move to the next lot with the arrow keys, or simply tap the next cell, and continue placing heights.
- Press Delete or Backspace to clear the selected lot if you want to replace an entry.
- Recheck the row and column you are filling so each line ends with one 1, one 2, one 3, and one 4 in some order.
- While the grid is still incomplete, press Check skyline or Enter any time for a non-penalty reminder of what is still missing.
- When all 16 lots are filled and you have reverified the edge clues from their own viewing direction, press Check skyline or Enter once to commit your answer; the first wrong commit costs one mistake and a second wrong commit deadlocks the run.
A useful habit is to treat the Check skyline button as a commitment step rather than a verification button. Use the rules in your head, or a separate sheet, for verification mid-solve, and reserve Check skyline for the moment the entire grid is intentional. This single mental switch keeps the incomplete check from looking like a mistake and stops the second completed check from ever firing.
What an Edge Clue Actually Counts
An edge clue is the number of buildings visible when you stand at that clue and look inward across the four-cell row or column. A building is visible only when it is taller than every building already seen, so the first building always counts and every later building must strictly exceed a running maximum to add to the total. The arithmetic sum of the row or column says nothing about the clue, because visibility depends on order rather than on which heights appear. A worked example makes this clear. Consider the sequence 2, 1, 4, 3 read left to right:
- Start the running maximum at 0; the first building is always counted because every height from 1 through 4 exceeds 0.
- Height 2 is the first cell, so the visible count becomes 1 and the running maximum is now 2.
- Height 1 is shorter than the current maximum of 2, so it is hidden and the count stays at 1.
- Height 4 is taller than the current maximum of 2, so it adds to the count, making it 2, and the running maximum becomes 4.
- Height 3 is shorter than the running maximum of 4, so it is hidden and the count remains 2.
- Result: this row contributes 2 to whichever edge clue faces it.
Because the clue depends on order rather than on which heights appear, a few clue values have shortcut meanings on a 4-cell line. The full shortcut map is summarized in the table below.
| Clue value on a 4-cell line | What it forces from that edge |
|---|---|
| 1 | The tallest building, height 4, must stand first; it hides every later cell. |
| 4 | Heights must rise strictly as 1, 2, 3, 4 from that edge. |
| 2 or 3 | Several different row orders produce the same visible count. |
| 0 (omitted, shown as a dot) | No extra condition; only the row and column permutation rules apply. |
On the published board the clue of 3 on the right edge of the fourth row is a strong starting anchor because it sharply limits the permutations of that row. A clue of 1 anywhere tells you the 4 sits in the lot closest to that clue. These shortcuts make the run tractable without ever bypassing the Latin-square logic that the rows and columns must also satisfy, and they explain why the omitted dots in the other eleven edge positions are quiet rather than necessary.
Why the Two-Mistake Policy Exists
The two-mistake policy is not arbitrary. The verification suite that the published board passes includes two independent checks beyond "is the grid solvable." The release test confirms that the five displayed clues produce exactly one complete grid, and a second test removes each shown clue one at a time and demands that the count rises above one. Every removal produces more than one valid grid, which proves that each of the five displayed clues contributes information needed for uniqueness. The five clues are therefore described as sufficient but nonredundant — they are the smallest set that pins down a single answer under the row and column permutation rules.
With five clues doing that much work, the run is built so that a careful solver has room to correct themselves on the first wrong try while still preventing endless reattempts on the same grid. Allowing one repair keeps the playing field fair to anyone who misreads a clue direction, slips a 2 where a 3 belongs, or forgets the right-edge versus left-edge distinction. Triggering a deadlock on the second wrong try then forces a Restart before the score is locked, so the 1,000-point award really does represent a solved grid under the stated rules rather than a long string of lucky repairs.
Reaching 1,000 Without Triggering a Deadlock
Reaching the 1,000-point score is straightforward once you treat the three check outcomes — incomplete, completed-correct, completed-wrong — as the only states that matter. Press Check skyline freely while the 16 lots still contain blanks, because that press is harmless and never uses a mistake. Save the single committed check for the moment the entire grid is intentional, after every row shows 1, 2, 3, 4 once each and every shown edge clue has been reverified from its own direction. If the committed check is wrong, repair the highlighted conflict immediately while the board is still editable; that single repair is the reason the two-mistake policy tolerates one slip. If a second committed check goes wrong, the run is locked and only Restart can reset the same empty board, the same five clues, the zero score, and the zero mistake count.
Keyboard players can keep their hands on the home row throughout. Arrow keys move between lots, digit keys 1 through 4 enter heights, Delete clears the selected lot, and Enter commits the Check skyline. The complete keyboard route is documented in Play Skyscrapers Puzzle on Mobile or Keyboard Only, which mirrors the same 1,000-point target from a controls-first angle. Pointer and touch players do the same operations through the on-screen number controls and the visible outline that marks the selected lot. For the broader Latin-square edge-clue definitions used in puzzle publishing, the public rules and terminology are documented on Simon Tatham's Portable Puzzle Collection — Towers. The 1,000-point award, the unique solution, and the two-mistake deadlock are properties of this specific fixture rather than universal claims about every Skyscrapers puzzle ever published, so treat the run as a self-contained solvable rather than a benchmark of any broader skill.
Related reading: Avoid Common Mistakes in the Spatial Memory Game.