That got me to thinking about adjustable game difficulty and what range of difficulties should be available. My theory is that different people require a varying amount of challenge to feel satisfied with a victory. Of course, everyone wants to win, but if the game makes it too easy for you, victory feels hollow. 'Too easy', though, varies greatly between players. One person might prefer almost always winning but still want to feel that they would lose if they played really badly. At the other extreme, someone might be happy with losing 99 out of a hundred games in exchange for the sheer rush of beating that one playthrough.
People are also differently good at actually playing games. For example, I am reasonably good at strategy games but surprisingly terrible at anything that involves timed jumps. Unlike most gamers of my approximate age, I didn't grow up with consoles that would have instilled these reflexes in me.
So, the question is: how often do you want to win? Defeats might loom larger in my mind than victories, but I think I like winning about half of the time. If I find myself winning 80% of the time, I will adjust the difficulty up a notch, and put it down a notch when losing 80% of the time. You might have quite different preferences. *
Adjusting game difficulty generally can't be done holistically. Only some game mechanics are amenable to being made more or less hard by adjusting, e.g., probabilities, accuracy, reaction time, number of steps thought ahead, et cetera. You will have designed your game mechanics to work together at the difficulty level you consider normal. Changing some mechanics to be balanced in a radically different way may stop them interacting properly with the rest of the game.
For example, if you are making a first person shooter, it's easy enough to make the computer enemies weaker: make them slow to react, give them bad aim, have them deal less damage or die more easily, reduce their numbers. But if you push this too far, the game becomes farcical: dull-witted monsters line up patiently to be shot, carefully aiming to miss the player. Even when it's more or less impossible to lose a fight, some players may still find the game too difficult. Why? Perhaps the real problem lies with the movement controls. Using WASD and the mouse to look at the same time may be something they don't have the coordination (or limbs) for, so the game just breaks down.
At the other end of the spectrum, I have a recurring pet peeve with turn-based strategy games. Picture this: it's late in the game and there are only three serious contenders left. The Republic of Me, the Empire of Computer and the People's Republic of Computer are locked in some sort of military, economic or scientific struggle. There is a fourth player, the Principality of Computer, whose territory borders on mine. The Principality lags far behind the three main players and has no chance of actually winning any more. Throughout the game I have enjoyed peaceful relations with them, perhaps a spot of trading. Then, suddenly, they send their armies to attack. The cities near the border are only lightly defended, because I did not expect the Principality to attack in such a foolhardy way. They take a few cities; I groan and aim the vast war machine of my country at their incursion. I spend emergency funds shoring up the defences and bring their advance to a halt. A few turns later, my army arrives, pushes them back and, since they're already there, wipes the Principality off the map. Shortly thereafter I lose the game, because this pointless war has caused me to fall irreversibly behind the other players.There are two general approaches to strategy game AI. It should either play to win or be a 'realistic' simulation of a government. The former would gleefully attack an ally if it saw the slightest advantage but the latter would not. The Principality's actions make no sense either way, because it can't win the war and our relations have always been friendly. What's going on? I'm playing the game on a high difficulty level. At this point, the only way to make the game hard enough is to make the computers irrationally hate the player—but this collapses a varied game about building, researching, diplomacy and war down into an inferior war game. I play war games. When I do, I don't mind that everyone is trying to kill everyone else, because the game has been built and balanced for this.
Outside the range in which gameplay elements have been designed to intermesh well, a game structurally changes and may no longer work. Getting the game to still function outside this range might require significant changes to its mechanics.
In conclusion, be aware that the range of difficulties where your game will work properly will probably be centred on the challenge-per-victory level at which you yourself play and enjoy the game. Your players will want to play at higher or lower difficulty settings, but providing the full range is not simply a matter of scaling up or down.
*It's easy to sit upon the perch of your own game skill and tolerance for losing, and sneer at anyone who wants their game to be easier than you do. This is stupid. Games are meant to be enjoyed. Someone might keep playing a game they have already mastered, because they enjoy the motions of playing it or because they want a dependable source of victories in their life. Games are (usually) fantasies of agency and it's pointless telling somebody they don't deserve that sense of agency because they're not skilled enough at playing.