Jump to content


  • Content Count

  • Joined

  • Last visited

Community Reputation

2 Neutral

Recent Profile Visitors

185 profile views
  1. On the Graphs : I don't know much about Desmos, but can you change the X axis or does it have to be the same scale of Y ? Just so we can see like from 0-1000 on Y and 0-10 on X in a normal computer frame. --- Otherwise, pretty cool. You could even introduce Magic armor of the target to calculate post-mitigation.
  2. This is simply your opinion, same as me when I say I think Blacksmith isn't fine. You can speak for yourself and I can speak for myself, but saying "they all love him" isn't remotely true. The fact is : in the previous forum, many voices were regularly heard (at least until about a year ago) how unfun it was to play against Blacksmith. So what we can say for sure is that "several people complained about him". And I don't intend to remove that at all. that's why I don't understand you're arguing against my proposition. Luck will 100% be there. Blacksmith will feel and behave functionnally the same. This is the current ability: I intend the post-modification Blacksmith to have THE EXACT SAME ABILITY DESCRIPTION, SAME ODDS. Everything! BlackSmith will have the same expected multicast, the same average damage, same max damage. Same odds of having a 4x multicast. What I propose is a statistical stability : lowering variance. That's it. And I won't go into more details without confirming you have at least some statistical background. Because I am in no mood to do a Statistics Crash course to explain my point.
  3. I'm not talking about a rework of Blacksmith. I'm not talking about any significant changes. I'm not talking about removing RNG. I'm not talking about removing "luck". I'm confused because nothing in your comment match what I said in the opening post. -- In simple terms : The same way the balance team deemed necessary to tweak Legio/Chronos with a counter because simply relying on an uniform distribution for RNG wasn't going to be the good solution, The same way DotA is working with Pseudo-RNG distribution, I'm suggesting to have a non-uniform distribution for BlackSmith too. That's all I'm saying. and I'm saying that the way the balance team solved the problem for Legio/Chronos isn't going to work for Blacksmith, hence he needs his own solution. A Normal distribution isn't less random than a Uniform. A Gamma, a Fisher, a LogNormal distribution are all as random as a uniform. It's simply not the same expected value and the same variance. That's it. All I'm suggesting is drawing number out of a non-uniform distribution. period.
  4. @ElementUser I get that. I since changed them. But it's still weird that our own posts trigger a notification warning. I had notification enabled for my post in Balance Forum for an email 1/day. On a given day, I did reply to a previous post. I was the only post in that thread for the day. I did receive an email notification at the end of the day "Hey Tartalacame, someone posted in your thread : it was Tartalacame!" Similarly, before I figured about the effect of "following" a thread, I was setup for "every post" notification. Well, again, if you post in such a thread, your own post will pop up a notification each time. I don't think that's an intended behavior.
  5. Is it possible to alter the "Follow Topic" email to not count our own posts ? When you follow a thread, your own post provoke an email sent to notify us that "someone has posted in a thread you are following". I'm like, yeah I know, it was me. And I'm not talking only about the "send notification every post" If I'm the only one that posted in a day, I will still receive a daily summary including only my own post.
  6. This is true. I fail to see how my suggestion affect that, but it is true that Blacksmith has an identity well defined, similar as some other key heroes, like Devourer, Flint and most first 30-ish heroes. The same way that altering Devourer's hook damage by 50 or giving/removing 30 units to Flint's range would change their identity, I fail to see how altering odds by an average of 0% and at maximum 10% would change Blacksmith's identity. If the status is hidden, they simply can't. If the status is publicly available, then you know your current level, but you can't "pump" up your numbers unless you have like 1 minute to spam your abilities and the mana to regenerate what you have lost. And even then, you cannot assure your future state will be better than your current, because should you have proc'd anything while you were "pumping" your "luck score", you suddenly step back to a worst level. The risk simply outweights the pros. So in both case, it won't happen (bare Blitz mode, but we don't balance for those either). It was something that happened only with Chronos with the auto-bash every 4 hit. But that was manageable since you could auto-attack a creep a few times to charge up just before a fight (knowing reliably that you won't proc) and then jump with a sure bash on the next hit. Here, you wouldn't cast fireball on a creep 10s before your fight "just in case" you won't proc, because if you do multicast, you lower your future chances. And even if you do not multicast and your odds raise up, you still wouldn't be sure your next spell will multicast either. That is true. It's different and I don't see it overlap, but it's worth briging up in the conversation.
  7. There is a difference between boosting 4x multicast odds and all multicast as a whole. We can play with the odds of "having a multicast" without touching the odds of "having a 4x multicast", especially the way the current coding is designed. The current code roll and if you meet a threshold, you have a multicast. Then a second roll is made to know if you have a 3x multicast, then a 3rd roll is made to know if you have a 4x multicast. We can play the odds of 2x multicasts and make sure the 3x and 4x multicast odds stay the same. That's what I was saying with my statement. Now, we can *also* play with the 3x,4x multicast odds, and that's part of the discussion I hope to bring here. And apart of the "I like chaotic RNG", what else do you want to say ? I've brought arguments about why I think a balanced RPG, altering the odds barely to bring more enjoyable plays, is a good thing, and why BlackSmith is different that Legionnaire/Chronos/Basher in that respect. Why their solution can't work on Blacksmith and why a more consistent RNG would be the better solution. I have failed to see your arguments.
  8. No, that's not how I suggest it should work. It has no impact on the odds of having a 4x or a 2x multicast. And since it would affects and counts all casts, not just the Fireball, you couldn't tailor it to a nuke. The explicit goal is to avoid these kind of scenarios.
  9. I'm not talking about gettings pseudo RNG like Legio or Chronos, where the % of non proc increase such that at some point, you are sure to proc. I meant it to boost or lessen the odds by, let's say, 10% on a base chance of 50%. Even if you have already be "very lucky", your proc chances would still be like 40%, far enough to do a 4x multicast. Similarly, on the other side, you'd still only have 60% to proc, so you could very well end up not getting any multicast. All I'm saying is to vary the odds during the games to tailor them to the expected value, so to diminish the overall high variance. You'll still be able not to proc, and still be able not to multicast. You'll be less likely to multicast 4x 5 times in a row, and less likely to *not* multicast at all 5 times in a row. Yes, I forgot you already have something similar in Tarot. Fair point.
  10. The perceived problem with RNG-based ability in general is that they are fine when taken as a whole, but any individual occurence gives a binary results : it either happened or it didn't. With Legionnaire's spin, Chronos' bash, basher, and similar effect that procs on attacks, it's generally fine. You do enough attacks in a game that even if you were "unlucky" for a certain string of events, when looking at the game as a whole, the sheer amount of attack will make it that you'll be close to the expected rate for the ability and generally people will feel that they met their expectetions. And even when it goes down to the "fight" level, you usually have enough attacks so that your ability will proc (especially given that most abilities have a hard cap and proc when a number of attack without triggers are met). So it seems fair. In the case of Blacksmith, the problem is that it triggers on ability. In a single skirmish, Blacksmith may have cast only 4 or 5 abilities. Even with 50% chances, on 5 cast, BlackSmith still have 18.75% chances to multicast 0 or 1 time. So maybe it only triggered once, and it was on "Frenzy", and Blacksmith was alone so it empowered a creep. The player feel like it was a waste. Or maybe it triggered twice (31.25%), but it was on the 2 Fireball he did cast. So the enemy on the other side felt very unlucky. Being "lucky" or "unlucky", and well within the limit of the RNG, may swing the fate of a fight (and maybe a game when late game) without any reason than "luck". I don't want to remove luck : I propose to reduce the high variance. Since there aren't so many abilities cast during a fight, and even during a game, I propose to tweak the odds to balance the fate. Without going too much into details (it's still the balance dump), I propose to have a hidden state on Blacksmith that has the "current luck status". This state would conceptually be a counter representing the multicast actual rates. High value would meant that the multicast have procc'd at a higher rate than expected and low value means it hasn't procc'd as expected. Numbers are to be tweaked and defined, but conceptually, it would be a "hidden buff" that would affect the multicast percentage by let's say up to +/- 10%. So let's assume for the sake of example that Blacksmith has a multicast chances of 50%. if he hasn't been lucky and the *actual* multicast in the last X spells (let's say 20) cast has been 5/20, (so actually 25%), the status would buff the multicast odds to 60%, and when it will procc, the current hidden status will update. So let's say after next spell he multicast, he is now at 6/20 (30%). Maybe the status will buff the multicast odds up by only 8%, to 58%. and it will readjust every time the spell is cast. When the average will reach 50%, the expected odds, the status will apply a modifier of 0%. If on the contrary BS gets lucky, and exceed 50%, the status will apply a "debuff" to the odds and lower them down, to a minimum of 40% (in our example). Now numbers could be tweaked, last X spells could actually be since the beginning of the game, or last X min, or last X spell cast. All to be defined (and it may also vary depending on how difficult it will be to code). Similarly, we need to decide if we carry over the "luck" when the muticast is leveled up, or do we start from scratch. And how will these % will vary with the "score" of multicast (do we consider a 4x multicast like 4 actual proc instances or just a single multicast occurence ?). I have my take on these questions, but I feel it should be openly discussed.A lso, I think the status of current luck should be hidden (to avoid the "let's prepare and buff the numbers before a fight), but a point could be made that it could be "public knowledge" too. So there is much left do discuss and defined, but I think it would make Blacksmith more consistant in his inconsistancy, more fun to play and more fun to play against.
  11. Well, I didn't look into the XML since about a year, so if it has been drastically changed, I can't say and will look again. You misunderstood me. I LOVE RNG. I want it to make it better, not remove it.
  12. Let's say I wanted to finally have a closure with BlackSmith RNG and propose a better, more robust RNG way to handle the Multicast, if I was to put the time to do it and submit a draft of the XML, would that be considered to be implemented ?
  13. Is there a way to reduce the spacing ? We can see much fewer posts in a single glance without scrolling (both in a thread and at the section/subsection levels). Similarly, the top menu + banner + subsection list is so "tall", I don't even see the first posts in a given section when I click on it. See Below:
  • Create New...