For example in 1871 there were 254 team games (col 3) of which 231 or 90.94% were complete games (cols 4-5). Teams used 1.09 pitchers per game on average over all games, but 2.00 pitchers per incomplete game or 2.00 pitchers given at least 2 pitchers (cols 6-7). (No team used three pitchers in a game during 1871-1873.) Expressed as percent of all games played by pitchers, there were 83.7% complete games (both start and close by one pitcher, col 9), 8.3% were other starts (col 10), 8.3% were other closes (also col 10), and 0.0% were middles (neither start nor close, col 12).

In principle, but not in 1871, there is a small overcount of closing relief appearances ("other close", col 10) and a small undercount of middle relief appearances (neither start nor close, col 11) because my desktop baseball database does not account for starting pitchers who leave that role but later return and close the game as pitchers.

In sharp contrast, by 2006 only 3% of team games were complete games for one pitcher (col 5) and 49% of all pitcher appearances were middle relief (neither start nor close, col 11).

Pitcher-roles.txt in the Files section of our yahoogroup website gives every series, 1871-2006, in one comma-delimited text file. 136 rows x 11 columns.
Pitcher-roles.txt (at yahoogroups.com, members only)

Testing: did Paul break it?

Bless any of you for not making me try to rank 64 guys at once.
I'll tell almost any of the breakdowns listed...

It seems to me that breaking down the pitchers so far as to have groups of less than 10 makes the rankings of those small groups neither especially meaningful nor especially interesting.

I know that we can't use the same formulas for the pre-1893 pitchers, but in terms of difficulty in ranking, is it harder to compare Charley Radbourn to Kid Nichols than it is to compare Rube Foster to Kid Nichols?</code>

Similarly, separating out the relief pitchers will preempt all the comments about how they are worse than the starters, but it will also tacitly reinforce that evaluation without putting it to the test. If even one relief pitcher finishes ahead of a starting pitcher in the rankings, that would be a highly significant event. I don't know if it will happen, but finding out will be a lot more interesting than not finding out.

Perhaps once we rank our SPs (w or w/o RPs) by era, it will be a lot easier for some of us to do a "top 20" all-time.
We have roughly top 20 for all other spots, granting that the pitching version would encompass not all of the HOMers.

Another possibility is picking the top 20 OFs combining the three positions, and even maybe the top 20 INFs if we liked.

Not sure on the latter two, but the pitching followup seems promising.

Of course, before we know it we'll be voting in 2009 HOMers (Rickey and a couple of pals), and I guess slotting them in at their respective positions...

is it harder to compare Charley Radbourn to Kid Nichols than it is to compare Rube Foster to Kid Nichols?

Yes. In my opinion, infinitely so.

Similarly, separating out the relief pitchers will preempt all the comments about how they are worse than the starters, but it will also tacitly reinforce that evaluation without putting it to the test.

Bill James broke them out in the first HBA, and only had 5 listed (and 30 RHSP) and I didn't think anything of it. Seemed like it made sense. It's a different position. We don't rank SS and LF together, even though they both hit and field. They have different responsibilities and expectations.

I like Howie's idea of ranking them in groups first, and then voting for a top 20 combining all groups. Would be best of both worlds, right?

Let me preface my responses to Joe by noting that the decision of how to divide up the pitchers will fall to Joe as commissioner: I'll be voting whatever the plan is. But I am not convinced that the divisions Joe is proposing make sense, and so I am trying to address what I see as the key problems with these arguments.

Yes. In my opinion, infinitely so.

Three points in response to this.

First: I'm sorry, but I don't buy this claim on its face. For Rube Foster, we have some very fragmentary data, we don't really know much about what his usage patterns were, we don't really know the level of competition he was facing in the data we have, and you claim that it is easier to rank him against another pitcher than it is to rank a player for whom we have fairly complete data and fairly good information about competition levels, just because a shorter pitching distance made it possible for pitchers to through proportionately more innings?

Second: I am also not convinced that the rate of pitcher attrition was significantly higher at the 1892-93 changeover than at any other time in the early game, once the contraction of 1892 is properly accounted for.

if one does a study, year-to-year, of the pitchers who threw at least 20% of team innings, looking at, say 1885-1901, which gives us eight years around the change-of-distance, does the attrition rate look significantly different? I would want to see those numbers before accepting the claim that the pitching distance change was that revolutionary a change. We see that the great pitchers of the 1880s all wrapped up their careers 1891-95, but that could be an accident of the random, uneven distribution of great players. To establish that this was a revolutionary change, there would need to be evidence of broader trends affecting the whole group of pitchers. Maybe that evidence is there, but we ought to see it before accepting that there was a radical, career-disrupting break here.

Third: Isn't it arguable that the role of the pitcher changed as much or more between 1875 and 1885 as it did between 1885 and 1895? If we are serious about comparing pitchers only in groups for whom "pitching meant the same thing," how could we toss Al Spalding in with the 1880s pitchers. If you look at the rate of Ks and BBs during his prime, it's radically different from the 1880s, much more so than the difference between the 1880s and 1890s. Why gloss over this difference while placing so much weight on the pitching distance change? Isn't switching from underhand to overhand delivery even more significant (not to mention other pitching distance changes -- see the argument that Tommy Bond's career was wrecked b/c of pitching distance change). If we can work through these issues to group Spalding with 1880s guys, because having a group with only one player is of pretty limited value for rankings, why can't we work through the issues separating 50-foot hurlers from 60-foot hurlers?

Furthermore:

I like Howie's idea of ranking them in groups first, and then voting for a top 20 combining all groups. Would be best of both worlds, right?

If the groups are not the same size, the only way to do this responsibly will be to have a free-for-all vote, which will require us to rank all the pitchers from all the eras together. If we are ultimately going to do that, why not just do that in the first place? Either we have the skill to rank pitchers from different eras against one another, or we don't.

If we do have that capability, then the division of the pitchers into groups is ultimately pragmatic, and should be done in ways that produce substantial ballots covering broad but still meaningful periods of baseballl history.

if we don't have that capability, and the division of pitchers into groups is necessary to produce meaningful rankings in the first place, then how could we combine them later?

Bottom line: it is inconsistent to ague that we absolutely must not rank pitchers from period A against pitchers from period B and then argue that that we can combine the lists to generate an all-time top 20.

Myself, I like Howie's idea, but I view the groupings as pragmatic, so I think we should create groups that will give us a good basis for going into a second round, which would mean groups of about equal size, that begin the work of integrating different periods, since our goal is ultimately to integrate them all.

Note that we're coming up on the regular 2009 HoM election (as in, will Rickey be unanimous?) - see Howie's post on the 2009 discussion thread. We shouldn't start the pitcher discussion until that election is over.

Oh, and Joe or John - can you fix the tags in Paul's #420, please?

Dan's post, not Howie's - sorry about that.

One thing I was thinking about... are we going to update the positional rankings to include the 2009 electees? If yes, how?