Wednesday, March 14, 2012

Diablo 3 projections

Sthenno came over today for one of our "shout numbers at each other" visits.  We talked a lot about Diablo 3 and how scaling will work in the game.  He is strongly of the opinion that having an achievable level cap is really poor design and I am inclined to agree with him.  Even given that we will get all of our skills and runes by level 60 I think there are some really good arguments for letting players level beyond that.  The only things players gain from levels in D3 are the ability to wear better gear, skills, runes, and stat points.  The stat point gains are pretty important at low levels but at high levels when we can expect to have stats in the 5,000 range (see next paragraph) so getting an additional 7 stat points from a level is pretty much irrelevant.  Certainly a brutal ramp up in XP cost to level would be appropriate; simply increasing the XP required to level up by 50% for every new level would allow really hardcore players to get into the late 60s and the most extreme lunatics could keep going into the low 70s but their advantages would still only be in the 1-2% range.  It would let people have fun gaining XP forever and have leaderboards for level but avoid any significant power inflation.

The other thing we talked about was how high stats are going to get.  Right now each class has a stat that gives them bonus damage.  The Witch Doctor, for example, gets % bonus damage equal to Int.  That is, if the WD has 100 Int they do 100% more damage.  There is a WD skill called Soul Harvest that grants 750 Int and heals for 20,760 on a 15 second cooldown when you are level 60 - unless they intend that this skill makes the WD by far the most powerful class we will need to have an Int score of ~5,000 at max level.  If we expect to receive 20,760 in healing every 15 seconds we must also assume we have a health pool of at least 50,000 which requires 5,000 Vitality.  Clearly my estimates are based on my gut feelings of what is 'appropriate' but either way I think we can be pretty sure our stats will sit somewhere between 3,000-10,000.  This has some fun consequences.

The biggest consequence I see is that other stats will also end up being this big as they scale in the same fashion.  Magic Find is the big one here - it seems clear that every character will be able to rock 1,000% MF without even trying and presumably people who really go for it will be able to have 5,000% MF or more.  If MF scales linearly then people would be getting 50 times as many magic items as 'normal'... which boggles the mind.  Given that MF scaled logarithmically (ish) in D2 and that getting 50 times as many items is ridiculous I am pretty sure we will see the same mechanic in D3.  Perhaps the first 100% works at full capacity, the second 100% works at half capacity, then 20%, 10%, etc.  Regardless of how the curve off works mathematically it is clear that after hitting 1,000% MF there will be virtually nothing to be gained by stacking more.

While it seems a little bit bizarre to have a system where absolutely everybody has enough MF that stacking more is completely useless it neatly solves a few problems.  First off it means that people won't really have to make significant sacrifices for MF on gear.  As long as you pick a bit of the low hanging fruit and get a little MF in the easiest ways possible you will have plenty.  Also this means that the furor over MF being shared amongst group members is completely irrelevant.  Everyone will have 1,000 MF anyway so if you are crazy and stack 5,000 MF for some reason you still lose practically nothing by giving it away!  Essentially this means that MF is a pretty meaningless stat as long as everyone gets the little bit that they need.  This reduces gearing to a performance based weighting and means that those with optimal MF gear won't really be able to get significantly more drops than everybody else - the best way to get gear is to play well and kill efficiently.


  1. Unrelated to the post but related to our conversation and probably of interest to anyone who is interested in the post:

    I tested Ray of Frost to see exactly what 215% of weapon damage means and it turns out you were correct - the damage is done over the interval of your weapon attack speed and the cost is paid every interval. Thus, if you use a speed 1.5 weapon you can Ray of Frost for around 5 seconds before running out of Arcane Power, and with a speed 1.0 weapon you can Ray of Frost for around 10 seconds. Wizards definitely want slow weapons.

  2. How did you come up with the 1000-5000 MF numbers? I looked at the items again the other day but they still don't have stats yet. Well, I guess the actual number doesn't matter too much, obviously you will be able to have "some" MF without sacrificing much killing power, or "a ton" of MF and less power.
    Another thing I'd been wondering since the new MF in party mechanic was announced: do you have to be close to your party members to share MF with them or can I buy a couple extra copies of the game and leave MF-wearing mules in town?

    Also, yay release date!

  3. The high MF number comes from two possibly wrong extrapolations:

    1) As Sky notes in the post, the amount of healing from high level abilities (as seen on the skill calculator) is so much that we know we will have to have a *lot* of health. Similarly, the ability that gives bonus Intellect presumably doesn't multiply intellect by two. If you only had 750 intellect then getting 750 would nearly double your damage. If you have 3000 then it's more like a 20% increase.

    2) On low level items magic find numbers are smaller than stat numbers, but not that much smaller.

    It could be very wrong for a couple of possible reasons:

    1) The numbers we've seen on the talent calculator, rather than representing level 60 values, just don't represent reality at all.

    2) Magic find scales up much more slowly than stats on high level items.

    I won't be surprised if you don't end up with that much, but I think it's very plausible that with some kind of Diablo 2 like scaling splitting your Magic Find four ways will be a lot more like losing 10-15% of your *actual* magic find.

  4. @Tom

    You have to be fairly close to other characters to share MF and XP. Basically you can't just have random toons with stacked MF in the game, they have to be playing with you.

    I guessed at the 5,000 MF number by noting that the amount of MF on gear is similarly proportioned to the amount of stats - an item with 9% MF and 9 Int would be typical. Thus if we have 5,000 Int we can assume that we can get 5,000 MF unless the scaling is not the same. I could be wrong about that scaling but I don't think I am.

  5. I really wish I could trust that Blizzard would be able to balance the numbers enough that 1000MF and 5000MF would be substantially different.

  6. As I said, there are good reasons to suspect that these MF projections may be off because our picture of level 60 is rather incomplete, and frankly 5000% MF sounds improbable. Also, I'm not sure that basing our guesses of how MF will scale off of 10-year-old Diablo 2 numbers makes sense because clearly Diablo 3 has quite a bit more thought put into the numbers.

    But all that being said, the reason they put and MF curve in Diablo 2 was because people got to the point that they could no longer get non-magic items. It happened largely because while Blizzard said Hell was going to be hard, it turned out there were builds that made it pretty easy so MF became the only stat worth thinking about.

    Mathematically the MF curve doesn't make any sense at all; MF gives you more items in a completely linear way which is multiplicative with killing speed, so there should be a natural point past which more MF is not worth the tradeoff. It's possible that MF behaves differently at high level than low level (because with more affixes per gear slot, you may run out of other affixes that multiplicatively increase killing power) but that should mean giving less MF per affix, not creating a curve where MF goes from overpowered to underpowered.

    Despite this, I think there will be a diminishing returns curve on MF and virtually any curve they apply is going to make it so that for highest achievable MF scores, the full score will not be very different from one quarter of the score. Not that there aren't curve that they *could* use that wouldn't do this, but the kind of diminishing returns curve that Blizzard seems to like is the kind where they choose the maximum value they want anyone to have, allow people to approach that rapidly, and then curve off to nothing.