I was particularly interested in his link to this post on The Wheel of Samsara, regarding the rarity of spells. The basic contention is that the whole of all knowledge that the Magic-Users have available is just the first level spell list. To learn something more, it must be found.
This post has a little to do with balance. In all my old school games, I've never had any sort of issue with wizards overpowering or overshadowing fighters, or any class seeming useless. I've had players that have held those beliefs of course, but upon reviewing actual play an individuals skill has a great deal more to do with success than with any numerical or class advantage handed to the player. (Indeed, in my experience these differences begin to disappear at high levels, rather than become more prominent.)
This has not been my experience in more modern games. Often I have found myself in games where a players 'build' (or my 'build') of a character overshadows the ability of the other party members. I've fallen into long discussions on other new school forums on the utility of differing spellcaster types, spellcasting versus combat, and various other discussions on how to avoid 'traps' in the 'builds' in order to just maintain your effectiveness (otherwise you might not be able to beat 'level-appropriate' encounters).
A small aside here; having played a more modern game and then moved back to my old school game, I've discovered a fierce reduction in talk about character builds (I need this feat, this power, and these class levels), and a proportionate increase in talk about actual play needed to improve characters. (I need to find some gold, and someone to sell me this, and a way to get a library)
There are several game factors in old-school games that keep the Magic-user from being able to survive on their own. Their experience table causes them to level slower than fighters, they have fewer hit points, their spells take time to cast (during which they may be interrupted) and they have no choice over which spells they learn.
I've had players ask me when looking through the book "Who would 'take' this spell, this other one is better". I point out that Magic-Users don't 'take' spells, they learn what they find. This is interesting in (at least) two ways. First, it insures not every magic user looks alike, and second it provides a lot of space for interesting spell design.
In my own game I actually hew pretty closely to the actual article, the only exception being that I give Magic-Users two randomly determined spells of the levels that they select upon gaining a level. We've been playing three months now in our current campaign, and our party conjurer just now reached level 3 last week.
In more civilized areas, I usually allow Magic-User spells of second and third level to be purchased. I only ask for 20,000-40,000 gp per spell for second level spells and 40,000-80,000 gp for third level spells. In most cases, they are better off tracking down materials and books towards an arcane library and attempting to research the spell themselves. They almost never have that amount of gold available.
It is these factors that are part of the definition of the class.
Another option in Dragon Magazine #17 by James M. Ward in the article A Wizard with a Difference, talks about specialty mages. These Magic-Users aren't organized by school, but by effect. They are very similar to Hackmasters sole practitioners. An interesting thing about this variant is that they may select to know any spell of any level on their list. Each spell has its own power level (1st-8th). He then memorizes two spells per experience level as normal off of his narrow and specialized list. There is a table of spell level 'chances'.
First Level : 100% Fifth Level : 40%
Second Level: 85% Sixth Level : 30%
Third Level : 75% Seventh Level: 20%
Fourth Level: 50% Eighth Level : 10%
They have a level *5%+spell level chance at successfully casting the spell, they forget the spell no matter if they succeed or fail. There are also many new spells contained within the article, including 'transport to the spot', 'heat ray', 'no punctures', 'hold sword' and more.
This is interesting because it holds parallels to the Vancian ideal of there being two 'levels' of spells, complicated and Byzantine. There are many downsides to it - it very much seems like the thinking that lead to some new school ideas. It's surprising how far back those ideas originated (1978!)
A post on that to come.