Our Conversation about Projects (continued)

[LA Bird]

A question we need to ask ourselves is how long we want this list to be. With 25 players in each of the 8 decades, that's a whopping 200 players total. At 2 days per round, that takes us to mid September 2026 at least. 10 players per decade is more feasible but that's still 80 total which is almost double the size of previous peak projects. And really, 10 players in each decade is not actually that many. Unless their entire prime fit exactly in one period, chances are many of the top 10 will also be in the top 10 for another neighboring period too. If we look at 56-65 and 66-75 for example, Russell/Wilt/Oscar/West will all be in both which means there are only 6 spots left for anyone else. Other than a few outlier peaks, we might just end up ranking seasons among the same group of top 30 players over and over.

So while I don't necessarily have a problem with players making multiple lists, I think the birthday idea would be better if we want to have a wider discussion over more players. As for the cutoff though, I would just go with simple decades. Best of the 20s (or 30s?) all the way to the 90s. A 69-80 grouping feels too jerrymandered and we would have to double check not only which year a player was born but also which bracket he falls into.

[One_and_Done]

As I mentioned earlier, if you want to do the project right it probably makes sense to hold off until after the top 100 project next year. This will risk bleeding into that, and I feel like alot of people will have better energy for it after they get a break.

[DraymondGold]

Like playing years better than birth years because we are comparing them as players. It's also a more natural way of thinking about players as we generally don't think about them by birth year unless we are comparing draft prospects.

If you do playing years, you end up with players on multiple lists - for example, LeBron would no doubt be on both 2006-2015 and 2015-2026, Shaq would no doubt be on both 1996-2005 and 1986-1995(since he led a team to the Finals in 1995), Jordan would no doubt be on both 1986-1995 and 1996-2005, Kareem on both 1966-1975 and 1976-1985, and so forth. By using birth years, you eliminate this issue. Each player appears on only one list, and this has the added benefit of opening more total spots across the lists for more players to be acknowledged, since you wouldn't have guys taking spots on multiple lists.

I think it also results in groupings of players that are more tightly/closely linked together. Like, for example, in a 2015-2026 list, both SGA/Luka and LeBron/Steph would be on it, and that's two different generations, IMO.

It'll be whatever the consensus is, but this feels cleaner to me, personally.

I think a significant problem with using birth years is that it requires a good bit of legwork from each voter, in order to actually determine who is eligible for each list. I don’t think any of us really know the birth years of many players, but we do know who was playing in what year. So each voter would probably have to basically backwards engineer their list by thinking about who was playing in a certain set of years that’s roughly 27ish years after the birth-year range being used and then check each player to ensure that their birth year actually makes them eligible. That feels a bit messy/frustrating in practice IMO, and could also lead to people forgetting to consider players that they’d otherwise not forget about.

I think I’d be in favor of doing it either (1) by decade, or (2) coming to some way of defining different eras of the league’s history (i.e. something like the eras that penbeast mentioned, such as pre-shot-clock, pre-ABA-with-shot-clock, etc.). If doing it by decade, it seems fine to me to have it end at 2025 (so, 2016-2025, 2006-2015, etc.), but using actual decades (so, 2000s, 2010s, etc.) also seems fine.

Personally, I kind of like the idea of using different eras the most, since it inherently minimizes comparisons between players who played in very different league contexts. The downside of it is just that it requires a lot more adjudication on the front end about exactly what those eras should be, and people probably won’t entirely agree on the era demarcation lines. But I think if the goal of the endeavor is in part to minimize discussion of era differences, then parsing things by era is the most appropriate IMO.

I’m personally not all that concerned about the fact that some players will show up in different eras. There are actually players who excelled in very different league contexts! For instance, LeBron’s 2009 and his 2020 are great years that genuinely came in different league contexts, and I don’t mind them showing up in different lists.

Just seconding jake's suggestion -- if the peaks project were to be split up into different time segments, I'd rather split things up into slightly longer time segments than a decade.

To me, decades just seem too small. Having traditional peak LeBron (12/13) in a different era from peak Curry, having peak Kobe (08) in a different era from peak Shaq/Garnett/Duncan (00/04/03), having peak Moses (83) in a different era from Magic/Bird (87/86), having peak Oscar (64) in a different era from peak Wilt (67) ... all makes it seem like a decade is too small a unit of comparison.

I also think the ordering of peaks within a decade would be less interesting, as the reduced timespan would reduce the number of outlier peaks, making it much clearer. For example, take 1976-1985 if we did decades going back from this latest season. The top peak would pretty clearly be peak Kareem (which wouldn't get too much debate), then Walton (maybe a slight debate with Erving/Moses but also fairly clear), then Erving / Moses (probably in that order but might flip)... and by this point we would probably be out of the standard top 50. There wouldn't be too many players at the top since a decade is short, and the order would be fairly obvious. What would the biggest debates be? Maybe we might get some interesting discussions with the great players who peaked in adjacent decades -- e.g. 84 Bird and 85 Magic weren't at their true peaks, but might lead to seem interesting discussions with Kareem/Walton. But again, the decade seems a little too small. And if we did 25 players per each of the 8 decades, that would likely lead to too many players to handle like LA Bird said above.

If we did go for comparing peaks within eras (not sure; up to mods/larger group discussion), I would go for larger eras. For example:
-we could do every 15 years (2011–2025 gets the pace and space era; 1996–2010 gets the dead ball era and the start of the freedom of movement; 1981–1995 gets the 80s and 90s dynasties; 1966–1980 gets the expansion era although does split the 60s players of awkwardly; 1950–1965 gets the start of the NBA in the 49-50 season from the NBL and goes through the early shot clock era).
-we could likewise do every 20 years (2006-2025 gets freedom of movement era; 1986-2005 gets the end 80s dynasties through the deadball era; 1966-1985 gets the expansion through the early merger era; 1947–1965 again gets the start of the NBA through the early shot clock era).

But I think it might be simplest and best to do ~25 year increments. It roughly lines up with the NBA at 50 and NBA at 75 (which are official NBA projects commemorating league history). It also roughly lines up with some major rule changes. Like Jake said, the project manager would have to specify the specific year, but there's a few clear options:
-Define eras by NBA year: NBA at 25 (1947–1976), NBA at 50 (1977–1996), NBA at 75+ (1997–2022/2025).
-Define eras by rule changes: NBA Pre-Merger (1947–1977) or NBA pre-three point line (1947–1979); NBA merger / NBA three point line through illegal defense rules / freedom of movement rules (1977 / 1980 – 2001 / 2004); illegal defense rules / freedom of movement rules through today (2002/2005 – present).
If we did by NBA year, that has the benefit of aligning with the official "NBA at XY" commemorative projects. If we did by rule changes, I might be in favor of using the NBA three point line and freedom of movement rules as the dividing rule changes (this makes the older eras slightly longer and the latest era slightly shorter; considering the new era will likely generate the most peaks, this might make for a more even split of peak players).

Pros of doing ~25 year increments over ~10 year increments:
-Three eras are manageable for people (so we don't get to ~200 candidates like LA Bird said). But it also gives us a nice grouping of players to compare within each era. E.g. in the 2022 RealGM peaks project, there were 14 players voted in that were during the NBA at 25 timespan, 12 players voted in during the NBA at 50 timespan; 24 players during NBA at 75+ timespan). So e.g. we could do top 20/25 players in each of the timespans, which might produce sensible results.
-There would be fewer occasions where the era boundaries divide up the peaks of players we often consider to be in the same era into different eras (e.g. unlike decades, peak LeBron and Curry are now in the same era; peak Moses and Bird/Magic are now in the same era; peak Oscar and Wilt are now in the same era).
-We still divide players that played in vastly different eras (e.g. no Russell vs Jokic) and divide controversial players who have been debated to death (Jordan/LeBron)
-And as above, there are more straightforward definitions for the eras, based on either official NBA commemorative projects (NBA at 25, 50, 75) or official NBA rule changes (NBA founding/shotclock; NBA merger / three point line; illegal defense rules / freedom of movement rules), which by chance happen fairly close in time.

There may be other ideas besides doing things by-era that work better (I haven't read through every post in this thread), but if we did a project by-era, taking larger timespans like every ~25 years makes the most sense to me.

[Doctor MJ]

Alright, I will wait for Doctor MJ to comment the idea of era peaks project. If there is no objections, we can start the thread on specifying eras.

I think Era Peaks sound great, and it would also mean that it would more a series of smaller scale sub-projects that people could pick and choose what they wanted to participate in, and we could schedule in breaks between sub-projects if we want.

I think it would make sense to start another thread for the project to further the planning and gauge interest more formally.

Regarding the epochs and splits, I'm glad this is getting debated and I'll put some thoughts in here while I'm on the subject:

* When I'm evaluating careers, I tend to like to use birth year or draft year as the primary organizer because each player only has one birth year, and in modern times, they only have one draft year.

When it comes to Peaks though, I'd rather focus on an explicit chunk of NBA seasons.

* Splitting by Decade makes total sense. It's not the only reasonable way to do it, but there's no fundamentally better way, and we group things by 10s naturally.

* In terms of where to so the actual splits, because we just ended the 2025 season, I think a 16-25, 06-15, 96-05, etc type of approach is the most natural thing, but I'll also say this:

The truth is that the "best of a decade" statement is true whether you're talking about 00-09, 06-15, 08-17, 03-12, etc, and so it can make sense to have overlapping epochs:

16-25
11-20
06-15
01-10
96-05
etc

* An immediate downside here is the fact that this does the opposite of the not-one-to-one mapping between player & epoch - it takes us to a point where we literally expect that some players are going to top more than one epoch.

* However, I think this is actually a good thing from the perspective of keeping us participants honestly with our logic. With non-overlapping epochs, it's easy to see a scenario where once a player "win" an epoch, people feel like they should give the nod to someone else in the next epoch even though we wouldn't want to defend the choice in a comparison outside of the project.

[Doctor MJ]

A question we need to ask ourselves is how long we want this list to be. With 25 players in each of the 8 decades, that's a whopping 200 players total. At 2 days per round, that takes us to mid September 2026 at least. 10 players per decade is more feasible but that's still 80 total which is almost double the size of previous peak projects. And really, 10 players in each decade is not actually that many. Unless their entire prime fit exactly in one period, chances are many of the top 10 will also be in the top 10 for another neighboring period too. If we look at 56-65 and 66-75 for example, Russell/Wilt/Oscar/West will all be in both which means there are only 6 spots left for anyone else. Other than a few outlier peaks, we might just end up ranking seasons among the same group of top 30 players over and over.

So while I don't necessarily have a problem with players making multiple lists, I think the birthday idea would be better if we want to have a wider discussion over more players. As for the cutoff though, I would just go with simple decades. Best of the 20s (or 30s?) all the way to the 90s. A 69-80 grouping feels too jerrymandered and we would have to double check not only which year a player was born but also which bracket he falls into.

Great practical points. I feel like 10 per decade makes a lot of sense

[Doctor MJ]

That's a good point. I can always adjust my thinking. If it's birth years, where are the best break points?

Well, I can just tell you how I'm doing it in my personal project.

(Acknowledging there's a whole generation of professional basketball players before this one that never played in the NBA, and for whom there isn't IMO enough data to do a meaningful ranking. Bobby McDermott and LeRoy Edwards seem like an easy 1-2 for that one, but I wouldn't really know where to go after that).

First generation is 1917-1929 - so this includes, Mikan, Arizin, Schayes, Cousy, Davies, et al. I stretched it back to 1917 really just to include Buddy Jeanette and Al Cervi towards the back of my list, as they were born that year. By stopping at 1929, it basically ensures that everyone in this generation(or at least those good enough to make a list) played at least part, if not most, of their career with no shot clock, while everyone in the following generation played all of their careers with a shot clock.

Second generation is 1930-1941. Pettit is one of the oldest in this group, but it's the generation of Russell, Wilt, West, Oscar, Baylor, etc. I stop at 1941 because I wanted to fit as many ABA guys as possible in the third group(so there's only few ABA guys - Zelmo, Hagen, maybe one or two others - in this group), and there are two many guys born in 42 that IMO didn't fit.

Third generation is 1942-1955. Connie Hawkins and Roger Brown(as well as Willis Reed) were born in 42. Stretched to 55 because that's when Moses was born.

Fourth generation is 1956-1968. Dream Teamers basically. Bird and Bernard King were born 56, Divac in 68.

Fifth generation if 1969-1980. Shawn Kemp and Larry Johnson born in 69, Pau Gasol and Yao born in 80. The generation of Shaq/Duncan/Kobe/Garnett/etc. Went back and forth on Kemp and Johnson but they were members of that 1994 Dream Team II where a lot of those guys were thought of as the beginning of a new gen.

Sixth generation is 1981-1993. Joe Johnson and Kirilenko born in 81, AD and Beal in 93. The generation of LeBron, Steph, etc.

Seventh generation if 1994-2005 or 2006(undecided). Basically, Giannis/Embiid/Randle were all born in 94(and Jokic in 95), so that seemed like a logical line. Undecided if Flagg represents the very end of this generation and the very beginning of a new, eighth gen.

Love that you're doing this personal project. I've got some similar stuff going into the deep pre-NBA past.

[Doctor MJ]

I like the decades idea because I feel like there are several decades I can comment on, and several I can't. The broader top-100 projects are really interesting and I've seen some historical games, I just don't feel like I have the ability to rank, say, Walt Bellamy versus Dwight Howard.

Again, if we do decades, make them year x6 to x5 so we can finish with 2025 (and because it matches up better to NBA eras).

Similar type deal would be redoing our recreating the Hall of Fame project which puts us in eras but without such rigid limits. (every 5 years, 5 people into Hall?).

I definitely thought of your Hall project from 2020 which I loved.

That project was based on when a guy's last season of play was, which is another thing like birth or draft makes a lot of sense if you're looking to have a guy slotted into one and only one cohort from womb to tomb.

But yea, for Peak, I'd rather focus on the seasons rather than biographical data.

[Doctor MJ]

This is a rough idea of what I came up with for the reverse order concept

1. Automatically include top 80 from last project
2. Voters list their top 20 players not already included
3. The 20 who receive the most mentions are added to make 100

Now we have our list of top 100 players but it is unordered.

4. The 5 players with fewest mentions from before are in the first voting pool.
5. Voters rank these players in order. The lowest vote getter "wins" the round and is #100.
6. Voters also list their 3 worst players to add to the voting pool next (no reasoning needed).
7. The player with most mention gets added to the voting pool and the process continues for #99.

It's a two stage system like Doctor MJ's but just in reverse. There is an extra initial list of 20 players but it's no different than an honorable mentions list at the end of the project (like this from the 2017 top 100).

Pros:
- Redistributes interest to later rounds whereas most voters would currently quit once their GOATs are voted in
- Opportunity to assess new posters and remove trolls/PBPs before they can vote in the top rounds
- Eliminates ghost votes because we have a complete list of players from the beginning
- Possible reward (tiebreakers?) to incentive participation in the bottom rounds

Cons:
- Like eminence said, there is a negative slant with a reverse order list. But by splitting into two stages for nominating and voting, I don't think the voting round that actually matters would be any different from past projects. The nomination round would be negative but it's just a list of names. Voters even now can go above and beyond that by trashing players they hate anytime.

So, LA Bird, I have faith in your ability to run this sort of a project logistically, but I don't have a good intuition for how it would go poster interaction-wise. I'd be inclined to say it might make sense to try with something on a smaller scale first before we use it for something intended to have 100 threads.

I actually think in theory Era Peaks could do something like this, but I don't want to shoehorn it in if it's not in the project runner's conception. If I were running that project, I expect I start with #1 unless my participants really pushed for me to change my mind.

[LA Bird]

So, LA Bird, I have faith in your ability to run this sort of a project logistically, but I don't have a good intuition for how it would go poster interaction-wise. I'd be inclined to say it might make sense to try with something on a smaller scale first before we use it for something intended to have 100 threads.

I actually think in theory Era Peaks could do something like this, but I don't want to shoehorn it in if it's not in the project runner's conception. If I were running that project, I expect I start with #1 unless my participants really pushed for me to change my mind.

Oh I agree. Many posters probably wouldn't like such a radical change and it's good to trial it first to iron out any potential kinks.

But I don't know if it's a good fit here for a new project like the era peaks anyway. As the first of its kind, we won't have the previous results needed to form the initial inclusions in the unordered list and each era is probably short enough to sustain interest without needing to go in reverse order. Unless the idea gets picked up by another project runner, the earliest opportunity to try it will probably be the next peaks project in 2027.

[lessthanjake]

A question we need to ask ourselves is how long we want this list to be. With 25 players in each of the 8 decades, that's a whopping 200 players total. At 2 days per round, that takes us to mid September 2026 at least. 10 players per decade is more feasible but that's still 80 total which is almost double the size of previous peak projects. And really, 10 players in each decade is not actually that many. Unless their entire prime fit exactly in one period, chances are many of the top 10 will also be in the top 10 for another neighboring period too. If we look at 56-65 and 66-75 for example, Russell/Wilt/Oscar/West will all be in both which means there are only 6 spots left for anyone else. Other than a few outlier peaks, we might just end up ranking seasons among the same group of top 30 players over and over.

So while I don't necessarily have a problem with players making multiple lists, I think the birthday idea would be better if we want to have a wider discussion over more players. As for the cutoff though, I would just go with simple decades. Best of the 20s (or 30s?) all the way to the 90s. A 69-80 grouping feels too jerrymandered and we would have to double check not only which year a player was born but also which bracket he falls into.

I think this identifies the biggest problem with the peaks-by-era idea. It will inherently involve a lot more rounds of voting, unless the number of peaks in each era is quite small. And if the number of peaks in each era is quite small, then it’d end up basically always being a discussion about the very top players ever and not much more.

One possible option here is to not operate this as a round-by-round project. Perhaps each era has its own thread, and people rank their top 10 or perhaps even top 20 or something just in that thread (along with explanations of it), and there could be some points system attributed to players for each placement (i.e. like 25 points for #1, 20 points for #2, 18 points for #3, 17 points for #4, etc.). Since it’d just be one thread for each era, the voting in each thread could be open for a long time and the project still wouldn’t last too long. That could hopefully allow time for discussion to percolate down to the lower-ranked players on peoples’ lists too. The positive of this idea is that it’s a way to do an era-by-era peaks project that allows for discussion of lots of players without the project lasting forever. So it gets around the problems identified above. The negative is that the discussion in a single thread wouldn’t necessarily be super concentrated, because people would be listing a lot of players in each thread. Another negative is that the investment to make one vote would be higher, though that’s counteracted by the fact that the overall time investment for participants would probably actually be lower (and therefore we’d be more likely to actually have people delving into the lower-ranked players than they typically do in a project that goes one by one down the line).