FunWiki: FlashNet

http://www.jurjans.lv/stuff/net/FreeNet.htm

A highly addictive Flash puzzle game consisting of a power source with one to three outlets, a series of "wires" of three shapes (L, I, and T), and a bunch of devices, all laid out on a rectangular grid. Essentially, route power from the power source, through all of the wires, to the devices by rotating the different squares of the grid. The game has a similar mentality to Minesweeper in that if you know that a certain square must have a particular orientation, then it follows that other grids must have known orientations as well. However, Net has more complex logic chains, no "death", and an annoying piano song that can fortunately be turned off. Also reminiscent of PipeDream?, but with no time pressure except for its influence on score.

There are two different forms of the game (discounting the grid size variations): non-wrapping and wrapping walls. In the former version, all of the walls are solid barriers, which makes the game for the most part not too difficult. The number of orientations you can lock down at the beginning of the game is fairly large (for example, if you ever see three devices, or two devices and a wall lined up next to each other, then you know the middle one points toward the interior of the grid). This version of the game, of course, nets (heh) you a lower HighScore. In the latter version, connections can be formed across the edges of the grid; essentially the grid is toroidal. Typically you only get a few locked orientations at the start and you need to use all of them, whereas in the non-wrapping form you can probably start anywhere with good success rates.

Sounds like a more sophisticated version of [NetWalk] from [Gamos].

I suppose if we like, we can put some standard deductions here. Might as well add some content to the page.

If you have a T wire next to a wall or a locked piece that does not connect to the T's square, then you know the T's orientation (i.e. away from that square). Simple.

If there is a large cluster of devices, usually there will be one or more whose plugs can only reach wires in one direction; this is a good place to start locking down pieces. (More plugs will become forced to point a certain direction as more pieces are locked down.)

Watch out for closed circuits (i.e. circuits that cannot connect to the power source). A simple example is two devices linked by an I wire. (An I wire directly between two devices cannot connect to either device.)

In fact, any contiguous area of I's must be parallel, since an I must connect and any adjacent I must either connect with it or connect with something else in a different square. (This requires the assumption of no "wasted"/unconnected pipes, but I have never had this assumption fail for large wrapped boards.) -AlexUtter

Has anyone noticed if the wires ever form loops? I don't think they do, but I'm not certain.
- They can. Make sure you don't have loops.
  - I meant in the solution form. I've definitely formed the wire into loops in the process of achieving a solution, but I don't think any of the final forms have had loops. I want to know if that's always true.
    - The ultimate solution will not have loops, at least for non-wrapping walls. With wrapping walls, I imagine you can, though. A good rule of thumb is: there will be no wasted pipe. I.e., no redundant pipe (loops) and no dead-ends. This covers the first point, as well as things like straight pipes must run parallel to solid walls. I believe this criterion and the 2nd point above are generally sufficient to fix all but half a dozen or so pipe segments... at least this has been true for me so far. As I said before, you cannot have any loops. The game will not finish if you have loops.
    - I've played games (on non-wrapping) where there were a few blocks that didn't appear relevant. One orientation interconnected them with current circuits, the other made them a cycle. The latter was not a valid solution (yes, I've checked).
      - Okay, a bit of thought has convinced me that there cannot be loops in any solution. Graph-theoretically, the solution is a tree, and thus has n nodes and n - 1 edges. Moving any edge to create a cycle therefore disconnects part of the graph (and thus that part does not have power). Any holes in that logic? (The setting of the walls does nothing to the graph-theoretic properties of the solution; it's purely an informational gain for the player).
A 2x2? alternating pattern of devices and Ts has two equivalent configurations:

  |      |
 -+--*  -+  *
         |  |
         |  |
  *--+-  *  +-
     |      |

Random question: How does scoring work? Taking 13:40 on a max-size grid with wrapping walls and making 107 moves when 106 were necessary (read: accidentally clicked on the wrong I-shaped wire) nets 449,926 points. Presumably making no mistakes is worth more than working quickly, but are more complicated grids worth more points? In other words, to achieve a maximal high score, must one find a grid where all 143 squares are out of position and then click through them in record time?

In an attempt to answer my own question, some results from simple (non-wrapping) 3x3? grids: 7 seconds, 7 turns needed, 7 turns performed, 90,000 points. 5 seconds, 5 turns needed, 5 turns performed, 90,000 points. Hm. 6 seconds, 6 turns needed, 5(!) turns performed, 90,000 points. 28 seconds, 7 turns needed, 7 turns performed, 56,250 points.
- From what I can tell, it's all about finishing quickly *and* accurately. If I get a good time, but make any mistakes, my score about halves on full size, wrapping. A bad time with no mistakes has a smaller impact on score, but it's still definitely significant. (As a side note: if memory serves me, the 1-million-point threshold is around 3:20 with no mistakes.) Also, on max screen size, it's very common to have an optimal score 1, if not 2, moves less than what is listed. These games feel like they're worth more, but I could just be on crack.--DanCicio
  - I would hypothesize that the game's minimum turns count is based on the the number of turns the game made to change from a valid solution to a scrambled puzzle. If this is true then the program assumes there is only one solution, which is evidently not true. Since multiple solutions are possible, it is probably not completely valid to assume that there can be no wasted connections, although they are evidently extremely rare in wrapping games. Anyone well-versed enough in GraphTheory? to prove anything about the game?
    - In every game in which I've tested a valid solution with wasted connections, it was not considered a win until I figured out how to "unwaste" said connections.
    - I just played through a game on non-wrapping walls in which, despite making two mistakes, I still finished in 99 turns out of a required 100. Wow.
      - That's nothing. I had a game where the minimum was listed as 98... and I finished in 92.--NateCappallo
        
        Actually, I just did a 3x3? wrapping game in which I finished in 5out of 8 possible turns. My ratio can beat up your ratio. Actually, the 7x7? and below wrapping games are all pretty tricky, since you often have very little information to start on.
    - The game's minimum turn-count is, I'd be willing to bet, based on how many pieces the game turned in order to create the starting layout. There are at least one and a half reasons why you can finish in fewer moves: 1) There are a couple of configurations that are not set, and one of them may take, e.g. one turn to reach, whereas the one the game started with takes three; 1.5) I don't think that the game checks to make sure that its "turns" actually changed state--if it flips an I piece end-for-end, I think it may still count as having been turned, but clearly you don't need to flip it twice more to get a valid solution.
      - Note that a "turn" is any number of rotations performed on a single device. This is how you avoid being penalized for taking a piece the long way around (e.g. rotating the piece through 270 degrees instead of 90). I do kinda wish that if you rotated the piece 90 + (-90) degrees then it wouldn't count against you; I make a lot of misclicks.
Finishing the puzzle in the minimum number of moves (or close to it) gives a huge bonus; with comparable times, it can make the difference between 100k and 800k points on the largest wrapping puzzle.
I begin to suspect that the number of moves required, as well as whether or not the walls wrap, determines your time limit, while the size of the grid determines your maximum possible score. Notice that the 3x3? grids have a maximum score of 90,000 (i.e. 3*3*10,000) regardless of how many moves are needed; if this formula is consistant then the maximum score is using a 13x11? grid (either type) and is 1,430,000. And having checked, the maximum score for a 5x3? grid is indeed 150,000. (It has additionally been verified that 250k is maximal for 5 by 5). Thus I declare the maximum score in this game to be 1,430,000. The difference with the wrapping walls is that you are given more time to make decisions with each piece. With non-wrapping walls, you have 1 second per piece on a 3x3? grid (and taking an extra second costs you in the range of 5k points; the clock is not as discrete as it seems), while with wrapping walls, you have...um...two seconds? I'm bad enough at wrapping wall puzzles that I can't maximize my score even on a 3x3? grid. Taking 3 extra seconds gave me a score of 63,000 or so, though. - ChainMaille
- Trying to do wrapped walls on a 3x3? grid is hell on wheels. You have no immediate information to use. It's a lot easier if you start with 5x5? or 5x7?. --LizzieKadison
The game doesn't take into account whether the walls are wrapped or not. It only cares about time and number of moves. You can get a much higher score with wrapped walls, because they can usually be done much faster. --LizzieKadison

I now present, for your edification and enlightenment, the scoring algorithm for the game.

Let n be 0 if you finish the game in the "minimum" number of moves or less. Otherwise, let it be the number of extra moves you required. Let x be the "minimum" number of moves. Let y be the number of seconds it took you to finish the puzzle, if its at least x, otherwise let it be x. Let M be the total number of squares in the grid. Then your score will be the greatest integer less than (1/(n+1)+x/y)*M*5000. At least, this works for every example I have tried; let me know if you find exceptions.