Someone want to explain Dr. Boom frequency?


You are also using your recolection rather than a hard data set.

Even then, unlikely events still occur. That dosen’t make them not random events.

Just means you need a bigger sample size.

At the end of the day, Dr. Boom hitting the field by turn 10 is likely to happen (more often than not it will be found by then, even more so when mulliganing for it)


play the deck yourself, you’ll be heavily disappointed to find out how little he shows up


I dusted most old cards, because wild is not my style. Then I bought “rise of shadow” cards that initially appeared to be enough to allow me to ignore having to play against cards from expansions that made me quit twice. Most insufferable are the mechs. I was very happy with my collection after buying the new “rise of shadow” cards, but it was not enough to ignore mechs. That feeling led to dusting everything. In total I have uninstalled three times this expansion because of the accumulation of awful design decisions, but reinstalled. Other card games do not offer the same level of aesthetic appeal. At this point I play to get in the last hurrahs before soon avoiding the card game genre entirely. So no, I won’t play the deck myself. I do not want to play with or against mechs. That expansion is the epitome of bad design imo.


Since all warriors hard mulligan for Dr. Boom it is almost a certainty that they will have it on 7 or 8. It is such a OP card that it can be held in hand from turn 1 and still win Warriors the game. Look at the stats on HSreplay — the rush from this card needs to go ASAP. It is completely OP.


Yep, without mulligan (at all), never drawing an extra card, and also going first, a warrior has a 43% chance to have Dr. Boom by turn 10.

Going second and mulligan hard for Dr. Boom? a 54% chance to have it by then without any additional card draws.

In either case, it is a card that you should expect to see in a warrior matchup, as by the time it would be hitting the field, it is pretty likely to be in hand (somewhere between 42 and 53% if no other card draw is used in the match).

One tough thing to grasp is that even rare events (such as seeing a Dr. Boom on turn 7-10 every game for a streak of 10 games) does not mean that the random card draw in the game is broken. Sometimes random events appear very not random. Sometimes rare, highly unlikely things occur. Someone often wins the Powerball lottery, even though the odds are around 1/300 million. Sometimes, you are just the lucky (unlucky) person to go through a string of 10-20 coin flips not working in your favor. A 50% chance of something happening does not mean that for every set of 10 games, half of them will have each outcome. In order to get that 50% chance through observation, you’d need a fairly large sample of games (100+, although you are really looking for more like 10k to FIRMLY say that it was a 50% chance) with accurately recorded draw data.

Long story short: The card draw mechanics are likely working just fine, you are either just very unlucky here, or misremembering, or a combination of both.


Dr Boom has the 3rd highest mulligan win rate.

The Warrior is actively looking for it, thus its early appearance is increased by mulligan choice.


It’s Call of the Wild all over again.


Truly random simply doesn’t exist when it comes to computers. The best rngs just hide the fact that they’re rngs very well and don’t get called-out.

HS has a very bad rng imo. You often get the same result doing the same thing: Like you play 2 matches in a row and a card x gives you the same random card into your hand both times. That’s statistically insane but happens to me all the time and let’s not talk about the amount of pack-duplicates (something everyone should have noticed).
Does that mean the rng is rigged? No. If this happens to every player (and I believe it does), we’re just talking about a very badly coded rng.

About Dr. Boom (as I posted earlier): Turn 7 is not that strange (considering how many cards both players already drew).


Standard counter intelligence answer. I get it. We do not know what we think we know. Well said. Did you know that Blizzard devs know that we do not want what we think we want? I know. This is all very confusing, yes. Do we know? Don’t we know? Does a person need 100 games or 10,000 games to be sure? Certainly, it must be a number so substantive that one person could not possibly do it, right?

Counter intelligence gibberish aside, I did gather hard numbers to learn how serious the manipulation was. Dr. Boom, is just one, of course, so let me share a little of this data using a random spot check of 11 games. The presumption is that manipulation that favors Dr. Boom draw exists. We should see that pattern develop in a small random sample. My one random spot check will suffice.

1st|2nd Drawn? Decksize Rate Draw Turn Play Turn Game End Turn
2nd T 13 70% 11 11 18
1st T 18 50% 8 9 32
1st T 27 20% 0 7 27
2nd T 22 60% 5 7 17
2nd F 9
2nd F 9
1st T 16 57% 7 8 13
2nd T 26 27% 0 7 10
2nd F 14
1st T 23 33% 4 7 13
2nd T 21 43% 4 6 12
  • NOTE: Only played cards can be verified, so it could be in hand with an “F”.

  • RATE FORMULA: (30 - Decksize + (3 or 4)) / 30; indicates average draw rate

Firstly, the game records indicated that players generally do not mulligan all cards for Dr. Boom, but it was accounted for in the rate formula to settle that argument. Next, we need to acknowledge that the crucial number in the table is the deck size. The play turn is of consequence, but it does not satisfy our need when developing facts. Subsequently, we need to acknowledge that we have not agreed upon a standard deck size that would equate to a turn ten play to meet our average expectation.

Now, what do we glean from this facts table? Well, the obvious first thing is that Dr. Boom is fairly consistently drawing within the first ten turns: 7 of 11 (~64%). One thing about statistics that I use as a personal rule when judging randomness is that a fractional value will somewhat frequently deviate by one fractional amount of +1/+1 or -1/-1 (without falling below a numerator of 1). Thus, 1/2 (50%) will somewhat frequently deviate to 2/3 (66%) or 1/3 (33%). This spot check shows close to that upper deviation, so it does not suggest anomaly. However, if repeated spot checks show the upper deviation, then the lack of normalcy should prompt a larger study to allow us to more ably scrutinize. Also, we need to remember that the 64% is really 64%+ due to the lack of data from the three games Dr. Boom was not played. Furthermore, the turn 11 play is literally one turn after the cutoff.

In conclusion, we should get more data. The first spot check leads to a need for more data due to the upper deviation meeting the presumption of draw favor. What we are seeking is counter factuals. Draw favor would prevent or inhibit deviation toward the lower draw rate of 33%. If spot checks never show the lower deviation, then we can briefly conclude that draw favor exists. In doing so we would need honest spot checks. Conversely, searching through larger serial data for windows of deviation rates that meet the lower deviation ought to require a higher spot range, say (2n +1). Eleven games is fine for honest spot checking, so 23 game windows should suffice when cherry picking larger serial data sets. However, if we had larger (unmanipulated) data sets, then using the whole set would be more honest.


I don’t intend to go deep into this discussion, but this caught my eye. Isn’t the draw turn kind of pointless data as we don’t know how much card draw was played before then? You have the deck size stat there as well, which I would have considered more relevant. And by that we have 7/11 only without any mulligan.

On the other hand, if we want to consider the draw turn, shouldn’t we evaluate the proximity of the draw to turn 7 (or 6 with coin)? The turn when it becomes a good draw.


I am glad you brought that up, friend, glad you mentioned the importance of drawing on curve. You see, this discussion is about the relevance of Dr. Boom; as you said. If AI.RNG causes Dr. Boom (or any card) to have a favored draw, then the curve strength of the card would be a special factor when engineering the algorithm that prompts the card to be drawn on or near curve. So, if we are to look at Dr. Boom draws on or before turn seven, its curve value as given by the card cost, then we see in the facts table that it drew 6 of 11 times (54.5%) of the time. That seems reasonable, right, or is it? What if every card had a ~50% draw rate at its curve value? What if Khadgar drew on or before turn two half the time? What if Zilliax reliably drew on turn five half the time? Do you see how this becomes an argument that excuses actual manipulation? Cards do not naturally draw on their curve value given true (or digitally averaged) RNG. The only thing that could cause cards to draw on curve is AI.RNG (a.k.a. rigged).

Thank you for your participation. I am so glad you brought this up.


Patterns don’t exist in small random samples, period.


Fractals. I win. Where is the prize?


Fractals and card draw odds are very, very different things.

In a fractal each part is strictly not an independent event. There is a defined pattern being repeated.

Card draw is not going to follow any pattern in the small scale that resembles the larger scale results.

You can not use a sample for 10-20 games to show that there is a problem with card draw odds.


See people always say 40% to have or 70% to have but its actually 100% chance to have isnt it?

Certain cards are ALWAYS there.


I haven’t played Warrior since the Warsong nerf so do Warriors hard mulligan for Boom?

Thinking about the play rate on curve, seems like it.


11 samples tells nothing except that you didn’t do enough work for it to be statistically meaningful. Especially when you didn’t show how you picked those 11 games.

A lot more data. If you want any kind of cherry picking you would need a sample size of 1000 to begin with.


LMAO! I am trying to help others find the truth. Please do not attempt to reintroduce doubt with the same tired old excuses.

Statisticians sometimes use sample sizes of ~150 for introductory polling of voting, and there is no pattern to voting; only association tendency. Repeating words to maintain doubt about conclusions is only used by counter intelligence, which the public learns via propaganda, usually in the form of media or advertisement. What you think you know about statistics requirements, as shown by your post, is not accurate.

Let me help you understand by remarking about human mental function. It was once proposed that the human brains is fundamentally a pattern matching system. That supposition is not true. Pattern matching is ancillary to a pattern recognition system, which is ancillary to a predictive system. Let me clear that up. The human brain is exquisitely tuned to predict events while using incomplete data. The absence of pattern data requires our brain to do most of its processing to fill in expected data. We use memory for recognized patterns, but for other missing data the brain uses different types of logic. One of the causes for people enjoying gambling or for being interested in RNG is that it triggers our brains to do what it does best. We are keenly aware when there is a pattern or an absence of a pattern. It is not imagination playing tricks on us. Our subconscious is vigorously at work doing super-server level calculations to justify what we are observing. Because of this, we do not need thousands of cases to tersely conclude that a pattern exists. Our lives depend upon our ability to correctly determine patterns [EDIT: by guessing without absolute proof, but also when no pattern exists or a pattern is being actively disrupted - antipattern behavior]. We avoid danger. We recognize food. We are aware when true randomness is manipulated.

I hope that from this short discussion you will have more confidence in your own conclusions for the remainder of your life. The same goes for anyone who reads this. Your mind was made to predict the future. That is your mechanical purpose, as much as you are your mental state. Trust yourself to anticipate future events. Do not let propagandists and counter intelligence agents defy your brilliant discovery of what is to come. This is true for everything you predict, which includes noticing predictable patterns in Hearthstone AI.RNG, like with matchmaking and card draws.

Be good to each other! Spread love and joy. Also, may RNGzus be with you.


That is exactly the problem. Because we are “wired” to spot/look for patterns we are extremely bad at determining random events, like card draw/shuffle etc. as we are always looking for a pattern, even when there isnt one.


No. That is a fallacy promoted by those who were selling the idea that our brains are pattern matching systems. Actually, we think just fine without observable patterns. In fact, what we call “creativity” is the brain’s ability to defy pattern behavior. In reality, we are most comfortable in an environment devoid of pattern or perceivable order. It is not chaos we seek either. We like options. The strictness of patterns is deadening to our senses. Environments dominated by order interrupt our sense of individuality, creativity, enjoyment of leisure, and sense of well being.

No, my friend, we are “wired” for self expression due to our sense of unique perspective. Live friend. Live fully. Do not let these ideas you speak of hold you back or steer you inward in an unhealthy way. We are not such simple minds. We are all more than what you inferred.