Guild Wars 2

Shufflepants.9785:

It is “I don’t know”, but it’s the kind of “I don’t know” that is almost certainly a “definitely not”.

Well, no. You have no evidence one way or another.

In fact the only kind of evidence (anecdotal) we have is contrary to your position. I’d wager the reason you’re getting negative response is because it’s obvious to most here that you’re just being contrary for the sake of proving a point that anecdotal evidence is unreliable – although true in itself, it doesn’t make the opposite false.

As I said: somehow a lot of people have trouble settling for “I don’t know.”

Shufflepants.9785:

it seems as though most people do not notice any problem or else there would be thousands of people posting on the forums complaining about it

Ah. Well, okay, then you’re just straight out wrong. Feel free to ask in global chat what people’s drop-rates are like.

Actually it’s a form of argument from ignorance. (Unlike the name says, it doesn’t mean you’re ignorant, just means it’s fallacious.)

If you had been around during the great Wi Flag debate in Asheron’s Call (I was) you would have seen the same argument. Those people that were affected by this “bug” in the rng complained vociferously that they were bugged. Those that were not affected by it came back with the same arguments you see here. “It is all rng”, “you do not understand random number generation”, “random is random” and so on. However someone at Turbine did finally investigate and discovered that there was a bug that weighted the random number generator against a certain segment of the population.

We can all agree that there is no way for us, the players, to gather the data required to prove or disprove a bugged rng in GW2. We would at least have to know what type of rng they are using first and foremost and we do not even know that. However what myself and many others would like would be some indication/response from ANET that they will at least acknowledge our concerns and have someone look at the code for their rng and assure us that it is truly random.

Turbine did not want to do this for the longest time in regards to the Wi Flag in Asheron’s Call but finally did so. ANET should at least do as much.

You can read about the Wi Flag in Asheron’s Call here:

http://asheron.wikia.com/wiki/Wi_Flag

Random is not always random.

I’m slowly coming to a sad realization that these sorts of bugs may be a bit too high-brow for ANet’s engineers to discover.

Deniara Devious.3948:

It is possible that the random number generator (RNG) of Guild Wars 2 is borked.

WARNING: Very Technical Stuff ahead.

I have worked in the games industry and also read the source code of open source game projects. It is not unusual to find game code where the RNG is implemented badly. Sometimes the programmers have used a good algorithms from e.g. Knuth’s The Art of Computer Programming (TAOCP). While this book has efficient algorithms, they have several times been botched by incompetence or mistake of the game programmers. The programmers sometimes do some “magic” stuff to the pseudo random number returned by the RNG and do not realize that you have be extremely careful not to botch it up. Sometimes the RNG implementation has minor programming mistake. It will give you random numbers, but their distribution is far from random.

A lot of programming languages, for example C, C++ and Java, are using linear congruentatial pseudo-random number generator.
http://en.wikipedia.org/wiki/Linear_congruential_generator

As these as inbuilt into the popular programming languages, it is not surprising that many games use them. Guild Wars 2 is probably written in C++ and it might use this as well. I really hope it doesn’t!

Linear Congruential Generator is very fast way to get pseudo-random numbers. Unfortunately the random numbers aren’t very random at all. Depending on the seed value it is possible to get degenerate cases, where every second number is even and every second is odd. The programmers then often do tricks like discarding the lowest bits to get more randomness. This is NOT a good. Instead they should use a better RNG, like Mersenne Twister.

You need a good expert proficient in both mathematics and programming to analyze whether the RNG is working as it should. I worked for an organization (yes, not a company ) which is making money game machines (slot machines, video pokers etc). They had a person, who was responsible to the true randomness of the RNG. It is vital for such organization to make sure that the numbers are truly random and nobody is able to predict them. Despite all this one game programmer had added one line of C code, which just limited the number of random numbers from a huge number of randoms into smaller set of numbers (very typical thing for programmers to do, to get random numbers of just certain numerical range). As result some numbers became 20% more common than others. Luckily our internal testing measures caught that error before that money game machine was placed in public.

Without access to GW2 source code we can just speculate whether their RNG is borked or working as it should.

PS. Over 2000 hours of GW2. All 8 professions played to level 80, except guardian, which is still level 62. Only few random exotic drops, no precursors (despite using the Mystic Forge with tons of level 80 rare weapons), 2 random ascended rings (1 from fractals, 1 from many hundred WXP chests). I have never gotten an item (dye, rune, sigil, exotic) worth even 10 gold. Every single drop in game has been worth less than 10 g.

i had thought about something like this a long time ago, less the technical parts.

i never had any precursor after 1000+ hours in. Threw 200g into zomorros too.

i thought it might have to do with the personal story traits that we choose at the start of character creation.

or this account is just plain unlucky.

I think there is a possibility that something is broken.

In my guild there are 3 people, who get the good loot. Well they only got one pre each, as far as I know, but lots of exos, ascended weapons, … If someone gets good loot in the guild it is one of them 90% of the time.

The only evidence I have that the rng is working is Anet saying it is. They been proven wrong before. So why should I take their word as proof? A: I shouldn’t.

Every day (almost as I have missed a few, but have 4k+ hrs 12+ AP) has eluded me to think that it is either broken, bugged, or just incorrectly implemented.

We all know those players who get everything and those that get relatively nothing. I saw someone give the example of a lottery. Well my reply to that is if someone consistently won the lotto, I can guarantee there’d be an investigation. If someone wins consistently at a casino, there will be trained eyes on them, examining everything, and lots of reviewing from the security room. These types of things scream of something not being right as does the rng in this game.

What evidence is there that it is working properly?

In all MMOs, I never had the best luck. In WoW, I was kicked out of 2 raiding guilds just because of my luck. I was the low man on the totem poll, and the gear I need wouldn’t drop for me in time for them to move on to the next tier of raiding, so they kicked me and took someone else already geared in my place. When I finally got into a guild that didn’t kick me, we raided the same raid for 3 months every week till I got my gear and then we all moved on together.

I literally feel like I am the most unlucky person for drops when it comes to MMOs. Probably why I like token systems for gear is a far far better system.

Shufflepants.9785:

Or, instead of invoking all kinds of ridiculous conspiracy theories, you could actually take the time to learn about modern random number generators, realize that only really old games used RNG’s that were so weak as to be game-able, understand that the developers aren’t out to get anyone because that would be defeating their goal of getting people to like their game, and take a course in statistics and realize that given the sheer number of people playing the game it is entirely expected that there are players who have not had a single exotic drop and other who have had 30 of them drop.
Have you people not heard of a normal distribution?! Just look at it some time. The people in the middle are the people experiencing the average rate of some event, but there are still outliers at both ends in the 0.001 percentile and people up in the 99.999 percentile. Just due to random chance and the size of the sample set, there will be people who get shafted and people who win the lottery just due to random chance even over the course of many hours played per person.
Say it with me now:
Random is Random!
If you need some one to blame, blame the universe and its physical laws for being or seeming probabilistic instead of fixed and determinate.

This is what I call the “sucker’s argument”. It is like the guy at a poker game and one guy gets Royal Flush five times in a row, and the sucker is like, “Oh my gosh, I can’t believe I am witnessing such incredible probability actually happening!”

He thinks he is smart, because he knows what the probability is. But he is actually stupid, because he doesn’t realize that when something that improbable happens it is much more likely that the game is rigged than that it just happened by random chance. Similarly, some people have won more than $1million in the lottery on more than one occasion. One such woman that I have read about happened to have a PhD in mathematics and was an expert in RNG programming (and the lottery she won used an electronic system to choose it’s numbers, it was some kind of scratch it system). They reprogrammed their RNG system after she won the second time.

But a sucker would say, “WOW, that is amazing that she won two times, what incredible luck and I should bend down and kiss the ground in thankfullness that I got to witness something in my lifetime when the odds of it happening to a single person were one in a trillion and the world will probably not see another example of such incredible luck in 1,000 years.”

I’m just making up numbers there of course, but the important point is that most of us have played many different games with rare loot drops based on RNG and this is the only game that I (and apparently most others) have seen where players consistently complain that the RNG is broken and not actually random. To not suspect something is messed up in the RNG programming in such a case is foolish.

Maybe if you have less than an educated layperson’s knowledge of programming and RNG you might not realize that there are many pitfall’s, difficulties, and intricacies when it comes to generating truly random numbers with computer systems and that it is easy for programmers to get it wrong or throw an accidental monkey wrench into the system (as was already described by others earlier in this thread)….

Zackie.8923:

or this account is just plain unlucky.

Something an account should never be. They’ve come forward in the past about errors (e.g. ecto rates), albeit hesitantly, so time shall tell in this case too. Hang in there and keep grinding/farming!

Essence Snow.3194:

The only evidence I have that the rng is working is Anet saying it is. They been proven wrong before. So why should I take their word as proof? A: I shouldn’t.

Every day (almost as I have missed a few, but have 4k+ hrs 12+ AP) has eluded me to think that it is either broken, bugged, or just incorrectly implemented.

We all know those players who get everything and those that get relatively nothing. I saw someone give the example of a lottery. Well my reply to that is if someone consistently won the lotto, I can guarantee there’d be an investigation. If someone wins consistently at a casino, there will be trained eyes on them, examining everything, and lots of reviewing from the security room. These types of things scream of something not being right as does the rng in this game.

What evidence is there that it is working properly?

I want one of you to address this posters point. I don’t question the fact that some people get a precursor, while others don’t (or other “good” loot). What raises suspicion to me is that the same people consistently get everything while others get nothing. Why is that?

Pay attention to people posting on forums, guild chat, zones etc. The people that get at least 1 precursor will almost always get another 2-3 or more. In some rarer cases you have people getting 10+ (with 100g or less to boot – or free).

Like the person above mentioned, if the same person were to win the lottery multiple times then it would raise suspicion. Yes, someone wins the lottery every so often but I don’t recall anyone ever winning it twice, let alone 6+ times. Yet in this game where a precursor would be akin to winning the lottery (odds wise), the SAME people CONSISTENTLY “win” the lotto.

Draco.2806:

Shufflepants.9785:

It is “I don’t know”, but it’s the kind of “I don’t know” that is almost certainly a “definitely not”.

Well, no. You have no evidence one way or another.

In fact the only kind of evidence (anecdotal) we have is contrary to your position. I’d wager the reason you’re getting negative response is because it’s obvious to most here that you’re just being contrary for the sake of proving a point that anecdotal evidence is unreliable – although true in itself, it doesn’t make the opposite false.

As I said: somehow a lot of people have trouble settling for “I don’t know.”

Shufflepants.9785:

it seems as though most people do not notice any problem or else there would be thousands of people posting on the forums complaining about it

Ah. Well, okay, then you’re just straight out wrong. Feel free to ask in global chat what people’s drop-rates are like.

Actually it’s a form of argument from ignorance. (Unlike the name says, it doesn’t mean you’re ignorant, just means it’s fallacious.)

But we don’t have zero evidence. We have the evidence that players are always thinking that there are problems (where problems are unintentionally low and not intentionally low but the players think they need to be higher) with drop rates in this game and many other games. There have been many threads over the past year about people thinking there has been a stealth nerf to ecto salvage rates because they and their friends feel like they are getting fewer than they used to. And almost always the players making the claims based on their “feelings” were wrong. Just based on that I would estimate the likelihood of any given drop rate bug turning out to be correct to be around 1/100 or even conservatively 1/10. So far, there has been only one patch that actually did cause such a stealth nerf to the rates. On that evidence alone if one sees a claim about broken drop rates based on a single player’s forum post, we should take it to be only a 10% chance at best that there really is a problem.

But that isn’t the only evidence we have.

We also have that psychological test after test has shown that people in general are terrible at judging probabilities in numerous situations for numerous reasons. Please see this list if you need examples: http://en.wikipedia.org/wiki/List_of_cognitive_biases
So, since we have some players getting what they feel is a lot of loot, some players getting kittenty loot, or some players (like me) who feel like they’re getting about average loot considering the intended drop rates, if there is a bug, it is subtle, and only affecting some players. Therefore, it is unlikely that a player who is affected would correctly notice. People in general would tend to not notice even when affected, and others would claim there to be a problem even when there is none.

Then there is the fact that actual drop rate bugs are rare. There are many different things that drop loot, many different drop tables and almost all of them drop loot correctly. I would estimate that at most only 1 in 1000 drop event types are bugged; where a drop event is a particular type of object in the game that generates loot and makes an RNG call, so splendid chests, each mob type, each mob type in each event, mystic forge gambling recipes, and a plethora of other things that all have their loot determined in individual and different ways . And in this particular case, the type of drop bug that the OP is describing is completely unheard of. So, personally I think the prior probability for a bug of this type is more like 1/1,000,000 or lower, but I’m trying to be conservative here, so lets stick with the 1/1000 estimate.

We have that the prior probability of any given drop event being bugged at 1/1000.
P( B ) = 0.001
We can estimate that we get a thread about broken drop rates on the general forum like every couple of days, so the probability of a drop rate bug report is roughly 1/3
P( R ) = 0.333
We have that the probability of a player reporting a drop rate bug when there is a bug is roughly 90%
P( R | B ) = 0.9

So, if we we would like to know the probability that there is a bug given that we have a report, we just apply bayes theorem.

P( B | R ) = (P( R | B ) * P( B ))/P( R ) ~= (0.9 * 0.001)/0.333 ~= 0.0027
or roughly 3 out of 1000 chance that there is a drop rate bug given this information.

Perhaps you think my prior probability estimates are off. You can argue about the prior probabilities if you want. Suppose you think that we only see a report of drop rate bugs once a month so that P( R ) = 1/30.
That would make P( B | R ) = (0.9 * 0.001)/0.033 = 0.027 or a 3% chance that the claimed bug is real.
Or instead you think there are far more bugs in the game, some ridiculous amount like 1 in 10 drop events are bugged. That would make P( B | R ) = (0.9*0.1)/0.333 = 0.27 or a 27% chance, still more likely that there is no bug.

And I was already being really conservative about an average player’s ability to spot this kind of bug based on feeling.

Personally I think the values for these things are closer to:
P( B ) = 1/1,000
P( R ) = 1/2
P( R | B ) = 1/100

Making P( B | R ) = 2*10^(-5) or 0.002% chance that this bug is actually there.

I am not just trying to be contrary. I keep hammering the point about anecdotal evidence being unreliable because I think it is key to determining the likelihood of this report. Because anecdotal evidence is unreliable we should disregard it, because my anecdotal experience is the exact opposite, that I feel like drop rates for myself and those around me are spot on to what ANet intended. But we should ignore that too. So we have no evidence about the current situation, but we do have evidence of past drop rate bugs and past drop rate bug reports, and based on those we should find it far more likely that this report is erroneous.

I don’t have any problem saying “I don’t know” when I don’t know to say the answer to a question that has some kind of unique answer like “What is the capital of Thailand?” or to questions to which I have no way to even estimate the probability because there is literally zero evidence one way or the other like “Are there parallel universes?”, but this is not one of them. We’ve seen people reporting about things like this before, we’ve seen that most of the time people don’t know what the hell they are talking about, and rarely there actually are bugs in the RNG that developers use the actual drop data to detect, confirm, and fix. For those things we can at least estimate. I am not completely certain, but I am like 99.998% certain.

As for your offer for me to go ask in map chat about people’s thoughts on drop rates, I’ve seen some people complain about drop rates and think they need to be raised from their current values, not complaining about thinking they are bugged. I’ve occasionally seen people complaining in map chat about ecto drop rates being bugged, but they usually get shouted down by everyone else in map chat saying “NO, RANDOM IS RANDOM!”.

I’m well aware of argument from ignorance. I am not claiming “I don’t know therefore it must not be bugged” I am saying “We can’t rely on a small amount of anecdotal evidence so we should rely on prior probabilities, and the prior probabilities indicate it is probably not a real bug, therefor not a real bug.”

It’s simple application of bayes theorem. But I guess if people knew how to apply bayes theorem we wouldn’t have everyone and their mother’s thinking that you shouldn’t switch or that switching is just as good as not switching in the monty hall problem.

Am I the only one you ever thought about RNG like I did for IV/EV training ?

Guess I played some games too much…

Theftwind.8976:

If you had been around during the great Wi Flag debate in Asheron’s Call (I was) you would have seen the same argument. Those people that were affected by this “bug” in the rng complained vociferously that they were bugged. Those that were not affected by it came back with the same arguments you see here. “It is all rng”, “you do not understand random number generation”, “random is random” and so on. However someone at Turbine did finally investigate and discovered that there was a bug that weighted the random number generator against a certain segment of the population.

We can all agree that there is no way for us, the players, to gather the data required to prove or disprove a bugged rng in GW2. We would at least have to know what type of rng they are using first and foremost and we do not even know that. However what myself and many others would like would be some indication/response from ANET that they will at least acknowledge our concerns and have someone look at the code for their rng and assure us that it is truly random.

Turbine did not want to do this for the longest time in regards to the Wi Flag in Asheron’s Call but finally did so. ANET should at least do as much.

You can read about the Wi Flag in Asheron’s Call here:

http://asheron.wikia.com/wiki/Wi_Flag

Random is not always random.

That was an interesting read and an interesting bug. I still find it highly unlikely that the OP of this thread has spotted such a bug.

Cant we just agree that even if rng is random that doesnt necessarily mean its fair. I know they say random is fair because statistically no one has more chance than someone else to get an item, however at times it is disappointing to see that throwing in 400g into the mystic forge yields no precursor for me, but for other 400g yields two. I dont mean to dispute the system and say it is incredibly flawed but (this is if it is not already in place) they need to implement a system in which one’s “luck” is bound to the account meaning that if you go to your 1000 enemy or something like that you will for sure get an exotic. The journey to that 1000 kill might not be filled with exotics or rares or it might be but the main point is that you have some sort of guarantee. You can already see the impacts of the rng system on the price of precursors.
Just me two cents

champ.7021:

Cant we just agree that even if rng is random that doesnt necessarily mean its fair.

Spot on, whether its fair or not almost doesn’t matter because as I think the OP argued even if the rng is fair the odds are so small that there will always be players who could play an astronomically long time before getting a drop. Thus there should be a way to mitigate this. That there wasn’t anything in the game at launch goes to anet’s complete inexperience in these matters.

I can agree that the current way to obtain precursors neither feels fair nor is it fun. Really your best bet right now is just saving up to buy one. But you just spent forever grinding for something that some one else just had fall into their laps. But I think this is why they plan on releasing other ways to obtain precursors like some kind of scavenger hunt or crafting recipe.

Back in my old MMO, City of Heroes, they actually fudged the RNG to reduce the chance of runs of bad luck on to hit rolls. If you had over a 90% chance to hit but missed, the next attack will guarantee to hit; 80-90% you can only miss twice in a row; 70-80% three times, etc.

I wish i’d had “RNG Luck” when i went to purchase this game.

Medieval.1679:

marnick.4305:

www.arena.net > there’s a job application page. If you’re really as good as you claim, that’s your approach angle. If not, you should post on the forums.

You don’t give source code examples and assume everyone believes you. On that complete lack of proof, but with an implied degree of authority you try to make us believe the RNG is botched. For trolls and haters, that would suffice. Unfortunately I do know what you’re talking about and I need a little bit of proof, not just hyperbole and hearsay.

Credentials do matter in this thread. Do you have linkedin?

And you are talking from what sort of authority?

The details that he has offered do actually check out (if you had bothered to read them – but going by your response we can all see that you haven’t). You on the other hand have offered nothing by way of firm rebuttal but have instead resorted to a rather weak form of character aspersion.

Please consider this constructive criticism as you are essentially doing the exact same thing as you accuse that poster of doing.

Maybe in future you should refrain from hitting the submit key until you actually have a clue what you are typing.

I read the post, but unlike OP don’t claim to post from an authoritative perspective. The post basically combined a few things he might have read on the internet with almost infantile lack of terminology whenever he lacked actual knowledge of the subject.

It’s not that this topic isn’t open for discussion, but the OP tries to claim authority where he has none. Unless he gives proper credentials, his post is as good as that of any other person on the internet: worthless.

That said, I don’t have any particular feeling about the RNG in GW2. It functions more or less the way I’d expect it to. In higher counts, clovers and ectos always give what’s to be expected.

I have been thinking to write a book about game design and one topic there I want to discuss is randomness in games.

We all know the world is not fair, but players do not play MMORPGs with the main motivation of it also being as unfair as possible. In fact most players play the game in order to see their character progress, get continually better, experience new content, beat other players, monsters and socialize with their friends. In GW2 you can reach the maximum level 80 pretty fast and obtain full exotic set to him/her as well. Getting a legendary however requires luck and grinding or very deep pockets.

It can be very aggravating for a player to spend literally hundreds of gold into the Mystic Forge and get nothing. It is possible to throw even 1000 g in there and get no precursor. I think this sucks. Same thing for chest rewards and random drops. Another way is to give player guaranteed drops or time gate them. I have an alternative solution, which is both random, but yet very fair system, giving each player same returns in the long run. Technical rant follows:

Instead of the traditional pseudo-random number generator each player has several randomized lists of numbers e.g. numbers 0-999 shuffled in random order (using e.g. Knuth’s algorithm / Fisher–Yates shuffle). Every time player has some random change to gain important loot e.g. drop from major event or chest, mystic forge etc. the game selects a new number from that list. Once the list is depleted, the game gives the player a new randomized list. Thus each and every player will have equal amount of big and small and medium numbers in the long run. Let’s say a chest has 1% chance to drop exotic and 0.1% chance to drop precursor. Each time the player opens the chest, he gets a new number from the chest. If the number is 0 he gets precursor, if the number is 1-9 he gets exotic, if the number if 10-999 he gets rare. After opening 1000 chests, each player has received same amount of exotics (9), precursor (1) and rare (990). Some players just get lucky much faster e.g. get it on their 1st attempt, while others require 1000 attempts, but will eventually get it.

Of course the randomized list could be bigger than 1000, as memory is really cheap. That would just be 4×1000, which is less than 4 kilobytes per player.

Well that will prevent both apparently lucky players as well as unlucky ones. That would work for loot in general or specific great loot in particular but if to many activities use that list then their still the chance of lucky and unlucky runs for a specific activity.

Deniara Devious.3948:

…

Instead of the traditional pseudo-random number generator each player has several randomized lists of numbers e.g. numbers 0-999 shuffled in random order (using e.g. Knuth’s algorithm / Fisher–Yates shuffle). Every time player has some random change to gain important loot e.g. drop from major event or chest, mystic forge etc. the game selects a new number from that list. Once the list is depleted, the game gives the player a new randomized list. Thus each and every player will have equal amount of big and small and medium numbers in the long run. Let’s say a chest has 1% chance to drop exotic and 0.1% chance to drop precursor. Each time the player opens the chest, he gets a new number from the chest. If the number is 0 he gets precursor, if the number is 1-9 he gets exotic, if the number if 10-999 he gets rare. After opening 1000 chests, each player has received same amount of exotics (9), precursor (1) and rare (990). Some players just get lucky much faster e.g. get it on their 1st attempt, while others require 1000 attempts, but will eventually get it.
…

While the method does offer some degree of inevitablity, the downside to this method is that once the “optimal” drop has been acheived there remains a gauranteed “glut” of useless junk that the player must deplete before getting another shot another optimal drop.

If this is known to the player, if the optimal drop was undesired (i.e. you got the junk of the best, which was objectively useless to that character), and the glut is sufficiently large, the desire to abandon the character may be preferable to working through the glut.

Resetting the list once the optimal drop has been acheived is a possible alternative, and has been discussed elsewhere. Ultimately, it’s just a matter of offering the player some form of inevitability to getting a certain drop. Some people feel this is a good thing, others don’t.

Edit: After a little thought one thing that you could consider adding on to the above method would be to perform a second, pure random roll (same list, but non-depleting), and take the better of the two. Thus continuing to deplete the list, offering some form of inevitability, and also not precluding the chance of “getting lucky” even if you’ve “used up all your luck”.

I’ve played over 2K hours I believe… probably 10 exotics TOTAL from drops since pre release… no precursors. 2 of my guildies are actually almost to double digit precursors that they DIDN’T buy or MF…. I almost cried when a friend got 2 Colossus and Zap in 10 hours… MF to get Zap, sold it for hammers, got Colossus, sold it, bought more hammers, Colossus….

There’s indeed a curious looting system in this game. Some people noticed what i also did: that if u stop play the game for a while, and u start again u will get better rewards for a short period right after. But if u constantly play it u have overall lower chances to get rare loot. It’s a strategy whose purpose is to motivate players , but it’s wrong since the one who spend more time in it deserves a good drop also.

VLNt.9862:

There’s indeed a curious looting system in this game. Some people noticed what i also did: that if u stop play the game for a while, and u start again u will get better rewards for a short period right after. But if u constantly play it u have overall lower chances to get rare loot. It’s a strategy whose purpose is to motivate players , but it’s wrong since the one who spend more time in it deserves a good drop also.

one guy said in LA that he gets better chances in the mf with lower level toons. He said he kept one at level 2 for months just to forge. That made my day