I appreciate your kind words,
PantherLotus, and thanks for bearing with the following gigantic wall of text (should you choose to read it).
PantherLotus said:
Now, if one wanted to refute the usefulness of PNB for the majority of all software, he or she would have a great point; it doesn't lend it self well to anything but the biggest selling titles. Still, I acknowledge your point and would say that PNB wasn't intended for that purpose.
I used 1k-selling software just so the effect was clear. The larger point was actually that PNB simply describes a relationship between hardware increases and software sales. It doesn't posit any mechanism(s) for the link, and without that it can't serve as any sort of explanation or result in a fruitful, trustworthy model. For example, how do you separate the effects in a multiple-release week? (Did
Devil May Cry 4 drive all the spike, or was some of it due to
Disgaea 3?, etc.) Or, how do you set the threshold for "biggest selling titles" in a non-arbitrary fashion? And so on.
Furthermore, restricting PNB to blockbusters causes new problems. The historical maximum hardware delta from week-to-week (apart from supply pulses) I believe is around 120k, from the first holiday seasons for Gamecube and DS Lite. This means that extremely successful software can't have a high PNB, even if it's pushing record amounts of new hardware. In such a record week
MGS4 would still only have a PNB of 25%;
DQVIII would only get to 5%! Only
less popular titles could climb higher, and again this raises the question of where you break off, and why.
PantherLotus said:
I'm suggesting that PNB indicates the relative impact of the game -- and this is something that you didn't mention at all -- in relation to the PS3 LTD of 2 million.
But it doesn't have much to do with hardware LTD, once installed base is past a particular point (which would be different for each title). As more people buy into a platform, the chances of owners being interested in any particular title goes up. If you're right that virtually everyone who wants
MGS4 already had a PS3, then it wouldn't have mattered much if Konami delayed it until the PS3 had a million more owners. Indeed, the PNB could even go up, if the baseline sales for PS3 went down. Again, my general point isn't that PNB is somehow artificial, just that because of this and other confusing factors, it can't carry the analytical freight you're trying to load it with.
PantherLotus said:
I suggest that when the % of game ownership is that high, but the PNB of a title is that low (13.9%), it means the market for that game (and 'those types' of games) is essentially saturated on that system.
I disagree, not in principle but quantitatively. Any strong version of this conclusion would require a world where consumers always immediately satisfied their desires, and their desires never changed over time.; weak versions become less and less interesting. I understand that models have to make generalizations in order to be of use; they free us from the "tyranny of the particular". I simply think that the number and extent of simplifications required to empower PNB as a reliable predictor are so large as to leave the model too divorced from reality. For example, I think it would be far harder than you imply to decide what "those types" of games would be. Even your point about high attach rates being worrisome, while generally correct, is subject to important exceptions.
PantherLotus said:
1. I do assume that all major titles behave similarly, and the data is on my side. All major titles follow the front loaded model where 90% of all copies (and direct hardware impact) is measurable with the first two weeks of it being on sale.
This is not true. Here's a random selection of titles, from all points of the sales continuum, both casual and hardcore, and from varied genres, platforms, and years.
Code:
Week 1+2
Game Lifetime % of LTD
Diamond/Pearl 5445228 37.10%
NSMB 5192102 22.73%
Brain Age 3640092 2.13%
DQVIII 3538860 79.02%
MKDS 3085574 12.45%
Wii Sports 2973940 8.49%
F. Red/L. Green 2824674 43.97%
FF X 2325215 85.29%
FF XII 2305990 89.24%
MHP2G 2239261 61.45%
SSBB 1639846 66.00%
GT4 1066749 77.25%
Samurai Warriors 1024253 78.57%
MGS2 798179 68.27%
GT4 Prologue 747692 39.80%
Ridge Racer V 611507 49.84%
Taiko Drum Master 578993 15.45%
Yakuza 2 572917 63.30%
DBZ Budokai 543312 72.07%
Ace Combat 4 357742 74.14%
ToLegendia 342779 85.46%
ToSymphonia GC 322779 72.48%
Soul Calibur II PS2 144948 70.67%
Silent Hill 2 124793 73.38%
Baten Kaitos 108615 59.31%
Battalion Wars 55565 54.97%
Frontloading is very real, but it's neither as severe nor as consistent as you say. The data clearly indicates a more turbulent and opaque situation than you claim (and that you need if the PNB as currently construed is to be meaningful). I think the most telling example might be the extreme difference in legs between
GT4 Prologue and
GT4. Perhaps that can be explained away; perhaps
all the ups and downs can be explained away. But the amount of post hoc argument needed seems vast, and with every auxiliary hypothesis your general model--and the PNB which depends on it--will become less compelling.
PantherLotus said:
I said it earlier, but comparing MGS4 to Wii Sports is nefarious at best. . . .Consistency is important, but comparing the title on the most successful console of all time (to date!) to one of the most disappointing follow-ups in the history of console gaming can't be good.
I don't think they're identical; nothing could be further from the truth. My point is that they're both successful, both beyond expectations--no matter what you claim,
no one on NeoGAF thought
MGS4 would do these numbers, and the people that came closest were called crazy.
Wii Sports is, of course, an even more unexpected super-hit for the ages. My point was that its status doesn't somehow make
MGS4's lesser success cause for alarm. PNB may tell us something, but you haven't shown how or why it could be a reliable indicator a game's overall impact, much less of overall platform health.
PantherLotus said:
5. Clearly PNB will drop over time, just as tie-ratio for each title drops over time. We must use each with caution and make only the most careful of conclusions with each.
Absolutely agreed. Your conclusions are not careful. You figured exactly one PNB, and ran from there.
PantherLotus said:
I take issue with the assumption/assertion that I ever, EVER post charts for any other reason that thinking people will find them interesting in combination with my own interest. I ask that one would NEVER assume that I not post a particular chart because I'm attempting to hide data or have since been proven incorrect. That goes against everything true sales-agers believe and is hurtful at best.
Please accept my apologies. I don't think you are hiding data (you posted an updated version of that chart in this very thread), nor do I think you're ashamed of it. But I think it's disingenuous to say you never made any claims with the chart and its title. You were simply mistaken about the sales curve
MK Wii would follow, which doesn't say anything about your character. Predictions are difficult, which is why I try to avoid making very many. Sometimes--as with the weekly hardware predictions--yours are very accurate indeed.
MK Wii's downward slope may yet change again, to echo the long tail of
WiiFit and more closely follow the path you expected. That possibility is yet another reason to doubt the PNB's efficacy as a predictive tool.
PantherLotus said:
2. That WiiFit is clearly driving hardware sales for the Wii. Difference between it and MGS4? The first is clearly the audience. The second would be the install base of each system when it was launched: When WiiFit went on sale, the Wii had a 4 millionish install base, and the Wii was averaging around 30k for the previous 5 weeks before it went on sale. Also, it was immediately before the end-of-year hardware spike. When MGS4 went on sale, the PS3 had a 2 millionish install bas, and the PS3 was averaging around 10k for the previous 13 weeks before it went on sale. And it went on sale in the middle of the summer.
First, as I pointed out above, past a certain point installed base has little effect on PNB. Second, I'm well aware that the games are different. That's my entire point. Different games, different reasons, different effects--nearly the same PNB.
PantherLotus said:
I didn't say that MGS4's success is reason for Sony to worry, nor that Sony should suddenly start worrying if they hadn't already been before. I was simply saying that using these numbers, it's obvious to me that the majority of all people that were ever going to buy the PS3 for MGS4 have already done so.
Because of frontloading, that statement is true for the majority of all software ever. It's so trivial I doubt you would've needed to defend or explicate it as you did. No, you really did claim more than that, you really did claim that the low PNB for this game strongly implied (guaranteed?) low PNB for future games, and therefore inevitably low sales of PS3 hardware. That conclusion seems justified to me (at least for now), but I really can't find the linkage in the middle. Step two is still all question marks.