"Local Control Rule:" Although the boss monster terrorizing the first city in the game is less powerful than the non-boss monsters that are only casual nuisances to cities later in the game, nobody from the first city every thinks of hiring a few mercenaries from the later cities to kill the monster (unless it's you).
Sometimes these posts are like macing toads in a beer stein. Every long-term player is perfectly familiar with class and level-based monster assignment ... some DMs spend months or even years carefully working out with an eyedropper just how many monsters to throw at a party whatever their given level, number, distribution of magic items, etcetera, etcetera. I was one of them.
Problem is, the logic is sound and compelling - for the same reason the algorithm occurs in so many films and film sequels. Equally matched opponents produce a more interesting dynamic. Even better is the situation being unequally matched by a marginal amount - with the players in the deficient position. Dickens knew that overcoming odds that are not in your favor wins over an audience. It also wins over a party. There's little pleasure in crushing fourteen orcs when you're 6th level. And 14 orcs are too many when you're first.
For all the hue and cry about how worlds need to be potentially dangerous, and how encounters shouldn't be tailored to parties, actually carrying through with that philosophy doesn't make a very good fit in a campaign. Running away gets awfully dull awfully fast. True, I wish now and then my players would have the good sense to back off ... but as things happen they tend to be pretty resilient. They're able to raise or replace their dead as necessary, as they wisely keep a war chest handy.
It brings me immense pleasure in the first few rounds of a battle to hear the party gnash their teeth in panic, saying "We're all dead!" and "Let's get the hell out of here." But then two or three members of the party will advance, and then the magic will come out and the party will discover they have a great deal more strength and power than they really know. Recently I tossed 45 flesh golems at them, manufactured in a steam punk contraption and released all at once. After the initial terror, the druid formed a wall of fire, hemming them in, and the 9th level mage got off a fireball. The player rolled unusually high, reducing a large number of the golems' hit points down into the teens, making them manageable for the fighters. Yes, I knocked them around some, but the party handled the matter just fine.
What's important is that they didn't think they would, at first. Their first reaction - dead wrong by the evidence - was that they were all going to die. Back when they were 4th level, I threw one flesh golem at them that rattled them pretty good. They still had that memory. So when they overcame the far, far greater number, it surprised them.
Now, would it have been a better encounter if I had made it 450 golems? Or if I had rolled a d100 and come up with 7? No. In fact, it wouldn't have been better or worse. It would have been merely what it was. Except ...
If the number had been 450, I would have had to run it differently in order to retain the drama of the situation. Of the 45 that I created, the party in fact only killed approximately 4/5ths of them. The others walked off into the woods, in different directions ... because my intention was to force the party to go looking for them before they could find civilization. If I had started with 450, I would have increased this number of disappearing golems, and more than the party would have had to be involved with hunting them down.
On the other hand, if I had thrown only 7, that wouldn't have been enough to make things very interesting. I would have had to create some other 'effect' to keep the party's interest. Such as having the golems' bodies inexplicably, upon death, increase in temperature - invoking some greater and less tangible danger. This was, in fact, where I left the party. That is precisely what is happening to the golems they killed. They don't know why.
My point is that DMs, to retain a sense of tension, MUST increase the algorithm progressively. The party can feel success, they can feel they are getting on top of the problem - but if you want them to keep coming back session after session, there's nothing better than the Perils of Pauline. One more seemingly insurmountable cliff-hanger, one more very slightly mis-matched situation, one greater non-boss monster to be overcome.
Don't let the party know it, however. They don't need to know they're being jerked around.
It is our nature to be bored with things we can already master. And disinterested in things that can never be mastered. But something right in the middle ... that's the holy grail, my friends. It may be disingenious, but as people keep telling me, the game is about fun.
Or at least it's about as many life-threatening circumstances as possible.