Yeah.
Basically, my view of Sobers is similar to C_Cs. I argued the points in a bit more depth in "the greats" thread, but...
Sobers is a better bowler than he would appear because of his average, for several reasons. First and foremost he is one of the few players for whom the idea that they could have done better in one discipline if they could have concentrated on it is genuinely true. Sobers was not just a batsman who bowled, he was a batsman who was a specialist bowler. Not only did he bowl 40 overs a test (which stands up well against any specialist), but he was also one of the most prolific wicket takers of his time. Indeed, I believe in the entire decade of the 1960s, only Graham McKenzie took more wickets. He played alongside some great bowlers, was comfortably the greatest batsman of his generation and arguably the second best of all time, and STILL led virtually the entire world in wicket taking. This is an utterly remarkable feat, that nobody else in history can even come remotely close to.
Given these credentials, obviously one would ask why his average is relatively high compared to other bowlers of his time. I think C_C found the major reasons for this, which are to do with simple fatigue and the fact that his multiple abilities with the ball and the weakness of his bowling support forced him to fill roles in the team that he was unsuited to at times. Sobers was certainly a better medium pacer than spin bowler, but was required to bowl spin because a) his team often had no other quality spin options, especially late in the game on pitches more inclined to deterioration than todays, and b) it was far less stressful on his body, for a guy who batted for hours on end, fielded first slip and captained the team, as well as bowling 40 overs a test. Sobers was a passable spin bowler, a very good medium pacer, a brilliant fielder and one of the greatest ever batsmen.
To compare him as a bowler to Jacques Kallis and claim that Kallis is superior because of his average is simply ludicrous, and shows an unwillingness to take anything into account beyond the raw numbers, and even those raw numbers in only the simplest possible fashion. Kallis was a good bowler early in his career, when he was a poorer batsman than he is today, and was effective as a genuine seam option for his captain, and his record early in his career reflects that. He was never however a prolific wicket taker. He filled in overs in a strong attack boasting guys like De Villiers, Donald, Pollock, McMillan and Kluesener, and if he didn't have a good day he simply didn't bowl. As his career has progressed, this has become the case more and more, where now he virtually never bowls unless it is in swing-friendly conditions against Zimbabwe. If Kallis was, next week in the second test, forced to open the bowling and bowl 20-30 overs in an innings, he would be slaughtered, and if he was forced to do it every match his average would soon skyrocket. Similarly, if Sobers had the luxury of bowling only when it suited, only his preferred style of bowling in the conditions, and only when he wasn't being played with ease, his average would sit comfortably below 30.
They are simply impossible to compare with averages alone.