How about this. Think of the mining process as being like a variation of the game Bingo. Let's call this variation proof-of-work (POW) Bingo. POW Bingo is almost a reverse of how Bingo is normally played.
Let's say that for each game of this reverse Bingo, a dozen numbers are dropped from the drum of numbers and all twelve numbers are shown all at once.
And then there are many tables in the Bingo hall, each with a huge pile of Bingo cards on it. There is one table used per Bingo game.
So when a game starts each person playing grabs a card from the pile and looks to see if the card contains a winning Bingo. If it doesn't, the player tosses the card aside and grabs another card. This is the "work" being performed ... searching for a winning Bingo card. There is likely more than one card in the pile that would win the Bingo game, but the race is on to be the first player to find a winning card for that game.
The player who does find a winning card yells BINGO, notarizes the batch of transactions in the bitcoin ledger (which have been accumulating since the beginning of the game), and asserts ownership of the prize (currently a 25 BTC reward).
All players then want to move on to another table to start playing the next Bingo game but knowing which table to go to is something pre-determined by the ID of the winning card at the game that just had the winning Bingo. There's a chance the player claiming to have a winning card had lied or was otherwise mistaken so each player will first want to confirm the validity of the card that supposedly is a winner. The checks performed are that the card truly does have a Bingo (based on the numbers that dropped from the drum), that the prize claimed is no more than the amount allowed to be claim for that particular game, and that each transaction in the batch of transactions is valid (funds had not previously been spent in prior batches).
If that all pans out then at a later time the player who called the Bingo is allowed to spend the bitcoin reward earned from that game.
After verifying the winning Bingo card each player then knows which table of cards to move on to next. At that next table is another drum of numbers. That drum too now has its twelve numbers dropped so the next game of this reverse Bingo can start.
After each Bingo game is completed one result is that there exists a trail going from the current table back to the previous table, and from there back to that game's previous table and so on all the way back to the very first Bingo game that was played.
So there is this chain of games that were played, each dependent on the results (discovery of a winning card) of the prior game. The current Bingo game always occurs at the "tip" of the chain, and that chain's length will include each table along the path back to the very first Bingo game played.
It is possible in a game for there to be more than one person to have a valid winning card. When that happens there are multiple calls of "Bingo" and these could be yelled out at about the same time. This introduces a conflict because there ultimately needs to be only one winner per game.
The approach to solving this conflict occurs in a unique way. When there are two Bingos that are each valid for the same game, some players will move on to the next table pointed to by one of the winning cards while other players will move on to a different table as that was the table pointed to by the other winning card.
Thus in that instance the path for the chain of successive games will appear to have an end with a split. This is referred to as a fork. Some amount of work (players looking for winning cards) will occur on one side of the fork and some other amount of work will occur on the other side of the fork.
Eventually a Bingo will be called on one of these sides or the other and the conflict is resolved as whichever side is the first with a winning Bingo ends up being the side which extends the path for the chain further. The rule observed by all players is to recognize the chain's path which includes the most number of tables as the path to follow. If a player finds that some other path (generally the other side of the current fork) becomes longer (i.e., has a higher number of successive Bingo games in the chain) then the player abandons the current path and joins the others at the game table at the tip of the path that has the longest chain.
For the most part, this longest chain will be the one in which the most work (sorting through Bingo cards) is performed. Luck can be the reason why one side of a fork happened to win even though it had fewer players than the other but overall the longest path ends up being the link of games in which the most amount of work had been performed.
Because there is money that is earned from the per-game award, word gets out and more people will arrive at the Bingo hall to help sift through the cards. As a result of a growing number of players (and thus, more work being performed) it starts taking less and less time with each game before a winning card is found.
To counter this, periodically the quantity of Bingo numbers that are dropped from the drum is decreased, from say a dozen numbers to just eleven numbers. This decrease makes it more difficult to find a winning card because it generally takes more work to find a winning Bingo card from the table's pile when only eleven numbers are dropped versus the the amount of work needed when there were twelve numbers dropped.
It really doesn't matter which set of numbers are dropped, because for each round, nobody has an advantage -- everyone has equal access to sift through each table's pile of Bingo cards.
The end result is that no matter how many people are sifting through the piles, roughly the same number of Bingo games are held each day (targeted to 144 games per day, with one game occurring every 10 minutes).
This approach is what ensures that there is no cheating that would cause a significantly greater reward to be issued and why more miners arriving won't cause currency inflation (i.e., a higher production of the per-game award).
This would be great for the "Simple English" Wikipedia. – Nate Eldredge – 2013-08-20T15:03:39.540
2.1M -> 21M. Great description, btw :) – Alex Yakunin – 2012-08-06T22:13:53.450