Wednesday, November 14, 2012

[GW2] Precursor Gambling


I feel for this guy.

On its own, a random number generator is not fair, moral, or kind. It is just random.

Which is probably why RNG makes for such terrible gameplay.

Much of GW2's endgame* focuses on the hunt for cosmetic improvements - legendary items with no stat bonus, but unique skins / animations. To keep these items rare, the precursor weapons to legendary items are low-chance results from mystic forge gambling.

Using mystic forge gambling to obtain precursors is a controversial mechanic (at best). The forums drama has been fairly intenseFortunately, Arena Net heard the complaints and has some (unfortunately vague) plans to make obtaining a precursor both fun and legendary.
UPDATE [11/15]: So far, they haven't provided an alternative path to get a precursor, just 'slightly increased the drop rate'.  Booo.

But before the system is thrown into the dustbin of poor implementation, I wanted to take a moment and examine one reason why using RNG to gate progression is a terrible idea: the distribution of cost.

Luck and Cost

Some players get a precursor on their attempt... while others throw thousands of items into the forge with no result. I want to visualize that difference.

From the last post, I have a reasonable expectation that four rare items have a 20% chance of returning an exotic. I'm going to guess that this exotic has either a 1%, 1.3%, 2% or 4% chance of being a precursor. And I'll check three different ways of getting precursors:
  • Plan 1: crafting rares, selling all exotic outcomes until a precursor is found
  • Plan 2: crafting rares, reforging all non-precursor exotics
  • Plan 3: crafting exotics, reforging all non-precursor exotics
Under these conditions, here is the estimated (average) cost of mystic forge gambling (in gold), using greatswords as a baseline, broken down the the chance for an exotic to be the precursor.**

1% 1.3% 2% 4%
Plan 1: 210.83 150.80 101.67 51.83
Plan 2: 278.88 210.91 141.47 71.36 
Plan 3: 693.11 524.63 346.16 174.87

These averages are (generally) reasonable. Under these assumptions, even at a 1% chance, a precursor is around a 200g investment, making it expensive, but attainable.

There are two things to note.
  1. Crafting exotics to obtain a precursor is three times as expensive as crafting rares. This doesn't necessarily mean that using exotics to obtain precursors is a bad idea. They might have 3x the chance of yielding precursors - its just that nobody has the data to find out. Fortunately, in terms of puzzling out why luck is a bad way to gate progression, this doesn't matter, so I'll focus on plan 1 & 2 from this point on.
  2. Given the current price of precursors... either the precursors are much rarer than I've assumed, or using the mystic forge may be profitable even at current prices.***

Luck and Extreme Outcomes

Average cost isn't the issue. Its the distribution of cost that is the problem.

And that is where luck hurts players.

Here, I've simulated the results of 5000 players using plan 1 to obtain a precursor from the mystic forge, varying the chance that an exotic is a precursor. I've plotted how much the luckiest 10% of players pay for a precursor by gambling, the unluckiest 10%, and two measures of the average player (mean and median).
As precursors become harder to obtain (moving to the left), the average cost of creating a precursor for the unluckiest 10% of players grows exponentially. For example, if an exotic has a 1% chance to be a precursor from the mystic forge, the unluckiest 10% spend an average of just over 800 gold to get a precursor, compared to around 200g for the average player.

And that is the problem with RNG: extreme outcomes.

Extremes are great if you're lucky, but horrible for the person throwing gold away.

A simple fix would be to institute a progressively increasing drop chance once a threshold has been exceeded. Great way to fix the discouraging luck issue.

Legendary design

There are better ways to design legendary journeys.

Both Shadowmourne and the Fangs of the Father questlines were fun, and helped add to the status of the item forged. Shadowmourne's components: reforged weapon of the biggest evil in the world, hardened with the pure blood of an old god, drenched in the souls of the enemy, and decorated with fragments of the throne of evil.

Randomly throwing weapons at a genie just can't compete with that. (And the player-invented lore behind legendaries is a reflection of that.)

The problem with RNG is that players don't earn the item. Which cheapens it. And it cheapens the journey. Because there is no journey, just a slot machine. Imagine Arthur winning Excalibur at a casino. Not epic.

It isn't fun from moment to moment. Its preparing to have fun.



* Stuff-to-do at 80.
** The choice of greatswords here doesn't shape the cost much - while the TP cost of precursors varies wildly, most weapons have approximately the same crafting cost. I'm also assuming that exotics are sold to market in Plan 1 for 1g.
*** Of course, if they're rarer, then some of the cheaper precursors are tremendous losses.

13 comments:

  1. Interesting post, thanks.
    Side note: do you know if magic find affects MF chances?

    Also what is the most cost effective rares to craft per weapon slot and how are you figuring the price for your example?

    ReplyDelete
    Replies
    1. Magic find only affects monster loot drops, as noted by the devs.
      https://forum-en.guildwars2.com/forum/game/dungeons/Magic-Find-Chests/first#post598159

      For the example, I pulled down price data from GW2 spidy and compared the cost of crafting weapons to buying lvl 75+ weapons. I used the cheapest alternative as my price baseline.

      As for what is currently the most cost effective, I haven't checked recently, but its typically carrion (yay inexpensive bones).

      Delete
    2. Thanks for the info.
      How confident are you with your 1-4% precursor rate? I've heard they might even be as low as .005%. Any data on this or dev notes?

      Delete
    3. A guess, nothing more. I'd love to have data on this, but because precursors are so rare, that isn't really possible.

      Like I note in the post, if precursors are much rarer than I've guessed, then using the mystic forge to obtain one is a terrible idea. If true chance an exotic obtained from the forge is a precursor is around .005%, then the average precursor costs 2000 gold to obtain. Since they typically sell for less than 500g...

      Also, just be clear, the 1-4% guess is the chance that an exotic obtained from the forge is a precursor. So, if you're using yellows and shooting for a precursor, the chance of getting a precursor from any particular attempt at using the forge would be 4% * 20% = 0.8%. Or if its 1%, then 1% * 20% = 0.2%.

      Delete
  2. I've spend 400g and... nothing. Enough.

    ReplyDelete
  3. Just to avoid betting all my horses based on faulty calculations from my end:
    Say rares cost 30s now and exotics sell for 3g each, that would put the average cost (using plan 1) at 48g (4%) to 183g (1%), right?

    ReplyDelete
    Replies
    1. Could you point out where my calculations failed then?

      For the 1% case:
      On average every 16 rares leads to an exotic (1 in 5 attempts succeeds, but every 4 fails gives you enough recources to try once more). With 1% chance at a precursor, you're looking at an average of 99 wrong exotics before you get the right one. So the calculation then becomes: 16*100*30s - 99*3g = 183g

      @Scavenger hunt: months away would be my guess. Afaik they are still in the phase where 1 developer is trying to figure out a system for this to work, which is admittedly a lot harder than it seems. AS the effort to get one should probably be at least in the same scale as map completion (assuming you then obtain an account bound precursor, if not, that's whole other can of worms you open up there). Problem then is: how do you not make this into another grind (cause shit will hit the fan if it the large majority of people perceive it as such)? Farming 250 of every rare material might very well take less time than map completion (I don't know), but I sure loathe that part A LOT more than map completion (which is actually relatively easy, mainly because it's quite enjoyable in itself...).

      Delete
    2. Yup, I made a mistake in my napkin math. Thats a whoops. The cost of the expected number of crafts (the median) is 183g, the unluckiest pay 424g. Removing earlier comment to avoid confusion of passers by.

      Delete
    3. *Unluckiest 10% pay over 424g.

      This assumes you're getting 3g / exotic after AH fees.

      Delete
  4. This comment has been removed by the author.

    ReplyDelete
  5. The slot machine provider novomatic have launched their online games implementation. The famous games like the Book of Ra, Dolphins pearl and many more can be played online. Live casino software

    ReplyDelete
  6. This is done to provide tourists and visitors of accommodations as well as gambling entertainment. Most casinos also live entertainment activities such as music concerts. There are around five thousand casinos in the world. Roulettes casino

    ReplyDelete
  7. If you want to multiply your money within a few minutes, and you have what it takes to put down all your savings at stake because you think that your luck can make you millions, then the business of online Bitcoin casino is exclusively for you. cryptocurrency gambling

    ReplyDelete