63 réponses à ce sujet

[MartialArtsSurfer]

There's over 100 possible outcomes
http://narida.pytalh...p_checklist.pdf

but just 34 possibilities in Gold/Black

What makes you think they used the same libraries? Or implemented non-buggy calls to such libraries?

Since you're in software development too, then you know code bugs where the calls to the memory registers are common source of bugs (pulling from the same stack or memory register or wrong memory register just from a typo, etc)

I don't do game development (I programmed Oracle/ASP/PHP database software).

Modifié par MartialArtsSurfer, 29 mars 2012 - 11:32 .

[vivanto]

MartialArtsSurfer wrote...
There's over 100 possible outcomes
http://narida.pytalh...p_checklist.pdf

Thanks for the link, was looking for it for a while.
I said weapons, not everything (especially not weapon mods as I got those in a different row in those packs, just posted the last two items because it was already a nightmare to read).

Then retconning my previous post, it's 28 possible items (premium packs: uncommon), with a chance of 21 rare. From those, I've already maxed 9 uncommon (not counting characters). So 19 with a bigger weight and 21 with a smaller.

MartialArtsSurfer wrote...

What makes you think they used the same libraries? Or implemented non-buggy calls to such libraries?

The fact they're using third party "libraries" even at a much higher level? (ut engine). You probably heard the old saying, if it ain't broke, don't fix it. It is a much lesser and more viable assumption than what you're trying with the evil plot of RNG.

MartialArtsSurfer wrote... 

Since you're in software development too, then you know code bugs where the calls to the memory registers are common source of bugs (pulling from the same stack or memory register or wrong memory register just from a typo, etc)

Typo, seriously? You're grasping straws. Besides, ever heard of managed code?

[JohnDoe]

sad, it devolved from analysis.. to 2 guys bickering on how some program library is better than other..

[Mysteryman2000]

You still get the same character cards even though you have the class at level 20 and even though you have all the appearance options at max.

[Father Alvito]

vivanto wrote...
You mean you know this because you didn't graduate in computer engineering?  

You're right.  My degrees are in political science and economics.  But I also write enough code (to avoid manually manipulating millions of observations) that I realize sometimes I write bad code that doesn't work as intended.  It seems likely that I'm not the only person that suffers from this problem, and showing that the task is simple in principle does not prove that the developer got it right.

So I'm with MartialArtsSurfer on this one.  I'm inclined to believe empirical evidence over hypotheses on something like this.

You might try to do the same, so you'll understand how annoying it is to just buy packs nonstop for 20 minutes and write them all down, hence why I just decided to **** it after the 13th and bought the rest without logging.

Did I say that you had to buy them all?  It would be better if one person conducted the experiment, but that's asking a lot.  It would be much easier to coordinate to build a database.  We just need some people to agree to buy them one at a time and post results, and a few people to buy them in bulk and post the results.

Surely someone that's more current on statistics than I am can take a look at the drop algorithm and tell us what kind of sample size we'd need in both groups to say something conclusive.  I'd imagine it's sizable, although I seem to remember that those types of tests require smaller samples than you'd expect.  The type of work I do always involves a single sample, so I don't recall the math and can't be bothered to dig up a statistics textbook to find out.

Modifié par Father Alvito, 29 mars 2012 - 03:51 .

[vivanto]

Father Alvito wrote...
Did I say that you had to buy them all?

No, you just basically said to give you numbers or STFU. Besides, I didn't make this for your sake, if you've paid attention, you would've deducted that I already made these logs long before this conversation.

It stops being a hypothesis the moment you consider computers products of science instead of magic boxes, while drawing ignorant conclusions based on an evidence that involves chance.

What you are doing is simply described here around 4:50.

[MartialArtsSurfer]

vivanto wrote...

MartialArtsSurfer wrote...
There's over 100 possible outcomes
http://narida.pytalh...p_checklist.pdf

Thanks for the link, was looking for it for a while.
I said weapons, not everything (especially not weapon mods as I got those in a different row in those packs, just posted the last two items because it was already a nightmare to read).

Then retconning my previous post, it's 28 possible items (premium packs: uncommon), with a chance of 21 rare. From those, I've already maxed 9 uncommon (not counting characters). So 19 with a bigger weight and 21 with a smaller.

MartialArtsSurfer wrote...
What makes you think they used the same libraries? Or implemented non-buggy calls to such libraries?

The fact they're using third party "libraries" even at a much higher level? (ut engine). You probably heard the old saying, if it ain't broke, don't fix it. It is a much lesser and more viable assumption than what you're trying with the evil plot of RNG.

MartialArtsSurfer wrote... 

Since you're in software development too, then you know code bugs where the calls to the memory registers are common source of bugs (pulling from the same stack or memory register or wrong memory register just from a typo, etc)

Typo, seriously? You're grasping straws. Besides, ever heard of managed code?

You're welcome.. I just want analysis.. from scientific investigation (which was why I got annoyed when some suggested it was superstituion) to see if buggy nor not.

I love Mass Efffect 3 & I think the Bioware team & programmers are a great team but even the best programmers in the world makes bugs,
especially since according to
EA insiders/former employees & newsmedia about lawsuits against EA by their employees, EA was/is infamous for
making their programmers work 80+ work weeks, 12 hour days late after midnight bleary eyed
during crunch time
so ya, making the wrong procedure call or pulling the wrong memory location register or call stack or typo late night programmiing is more than possible, especially since Multiplayer was a LAST MINUTE RUSH JOB under time pressure to ship:

What if the RNG seed is just the seconds or minutes of the server? by design or bug instead of milliseconds?

1) THey're using the Unreal 3 Engine, it's about 5 years old -- shoehorning 5 yr old software code into a new program is unrelated to it is always going to require debugging & modification

Especially since ORIGIN, EA STORE & RNG unlocks & casino lottery is TOTALLY UNRELATED to the Unreal Engine..

they didn't reuse old code that was working.. they wrote NEW CODE FOR THE EA STORE & CASINO & tried to shoehorn into the Mass Effect 3 engine modified from the Unreal 3 engine already

Even porting their OWN CODE from Single-player to Multiplayer resulted in "simple" bugs:

Look at some simple known bugs that still haven't been fixed after 2 months ever since the demo  & release date:

1) Ultralight Mod doesn't work in Multiplayer but works in single-player. It shows cooldown in weapons screen but actually has no effect in game-play nor the powers screen

If we did as you suggest & not question the code's bugginess & never tested it out in actual gameplay & just took things at face value & assumed it was working as intended, we've never would've know it wasn't actually working despite the game telling us it was working on the weapons screen

The fact that it works in Single-player but not in Multiplayer shows that even something as simple reusing the Single-player code, someone made a typo/mistake & either forgot to enable the register or procedure call in Multiplayer or set the wrong flag when porting Single-player to Multiplayer....

so even PORTING THEIR OWN CODE is buggy, let alone 3rd party libraries that are 5-10+ yrs old

2) Drell Reave sound bug every every player hears the annoying Reave sound almost non-stop instead of just the person using reave --Bioware found & will fix this bug in the next patch.. after 2 months..... porting single-player to multiplayer

3) NON-vanguard Players & mobs falling through the floor & teleporting back to the landing zone or staying stuck in the floor or sky

4) Vanguards charging off the map or getting stuck in floors & maps

5) Mobs stuck or frozen in plain view right in front of you face but inert, but alive & immune to damage, requiring wave restart

So if they're getting bugs porting their own code from Single-player to Multiplayer from 5+ yr old software Unreal Engine library, what makes you think their brand new code for Origin, EA Store, & their casino is 100% bug-free? 

P.S. I like the Geth video you linked.. it shows also that people will ACCEPT THINGS ON FAITH ("EA Store, Casino & their software libraries are bug-free because they're just porting & using old code & shoe-horning it into new code")

I DON'T KNOW if it's buggy or not, but I do want to find out by gathering evidence/data..

but if you just assume ON FAITH that it's working fine, then we'll never find out.

 as a science person, you have to assume the hypthosesis that it's repeats are POSSIBLY buggy (or is it intended?) & experiment/gather data from there
instead of just saying,
"don't question it, it's fine, nothing to see here folks,
move along,
these aren't the droids you're looking for" LOL 

Modifié par MartialArtsSurfer, 30 mars 2012 - 08:13 .

[MartialArtsSurfer]

I just thought of & remembered something:

"It's not a bug, it's a feature!" LOL

What if the common repeats of the packs is a shortcut/intended feature to save processor cycles/memory usage from overloaded servers because, as much as I love Mass Effect 3 as one of the best games ever & think Bioware is one of the best gaming companies around:

1) EA reported Mass Effect 3 sold over 1 MILLION copies & shipped 3.5 MILLION copies in just the FIRST WEEK.. it's gone up even more in the past 3 weeks

2) With millions of players trying to connect to the EA servers/store, it bogs down & slows the servers (or server powered by a gerbil running on treadmill),--notice the delay when accessing the EA store & delay after buying a pack?

Resulting in the common disconnects & store errors due to the server(s) being overloaded (just like you get 404 errors when trying to access a server that has too many visitors accessing it)

3) To cut down operating costs & increase profits, instead spending millions on more servers/bandwith, EA tries to save processor cycles, time, & memory usage so
INSTEAD of initiating & creating new Random Number Generator calls for each request from millions of players every second/minute, it instead just generates ONE Random Number seed every few minutes
or
few seconds & uses that same number for the players and/or multiplies it by the players ID number or whatever
--this is faster & saves millions of random number generator function calls, memory registers, & processor cycles & reduces the need for more new servers

A news report showed that Google has over 200,000 servers & can afford it. EA doesnt'.. in the interests of maximizing profit & cutting costs, is the above possible?

Hmm, Day 1 paid downloadable content & a pay-to-play casino EA store & pay for an Online Pass if you want to play online if you bought it used shows that EA is primarily profit motivated like most other public corporations where shareholder dividends & stock price matters most, which is fine & the name of the game in business --so I don't hold any illusions over it

Modifié par MartialArtsSurfer, 30 mars 2012 - 03:45 .

[Fieryeel]

Ravenear wrote...

My idea to get better items is:
1. buy Recruit Pack until you will get all basic weapons on X level - on Veteran Pack's & Spectre Pack's you will not get crapy unlocks of basic weapons; it's fast step.
2, buy Veteran Pack (Premium Veteran Pack) until you will get all uncommon weapons on X level & weapon modifications on V level - right now im on this step i have unlocked all uncommon weapons on X level and trying to unlock all uncommon weapon modifications on V level
This is my way

Through this method, I 've gotten all the common weps up to X.

Do I continue buying recruit packs to get all the common wep mods, or do I move on to Veteran Pack to get the uncommon guns?

[Humming]

Since our team is not fully synchronized, we mostly play Silver. This means I can only afford one spectre pack for every two games I play. So with lobby getting ready and all; say I can buy one spectre pack once every hour with 20 min per game.

Yesterday I bought 5-6 packs. Got 3 Krogan Sentinel upgrades and 2 Revenants. With one hour between every purchase. I seriously doubt that it is a random number generator error.

I think this is just pure randomness and the human brain trying to find patterns where there is none. Just look at all the pictures of Jesus on a bread of toast that was popular a couple of years ago.

[Father Alvito]

I was in the shower (the incubator of all good ideas) earlier, and I suddenly realized that this is really easy to test.

If the server truly assigns drops in a non-random fashion when packs are bought together, then consumables slots should also be non-random. Presumably there is a 1/4 chance of getting any given consumable in any given pack slot (this would need to be verified). All we'd need is a sufficiently large sample of simultaneous purchases to test the hypothesis, which shouldn't be hard due to the limited number of possible outcomes.

Just as importantly, the experiment would be very easy to replicate, so we could quickly figure out whether or not someone is probably presenting biased data or whether a given result is simply in the 5%/1% of outcomes expected due to random chance.

vivanto wrote...

It stops being a hypothesis the moment you consider computers products of science instead of magic boxes, while drawing ignorant conclusions based on an evidence that involves chance.

What is happening here is that you are making a lot of assumptions that are, quite frankly, stupid.  Computers are products of science.  The humans that write the code are not.  If you can find the piece of code in question that generates the drops and demonstrate that it cannot possibly behave non-randomly, I'd call that conclusive.  In the absence of that, I'd expect to see empirical evidence before concluding anything with certainty about the contents of that piece of code.

Modifié par Father Alvito, 30 mars 2012 - 02:50 .

[Sypher90]

MartialArtsSurfer wrote...

What if the RNG seed is just the seconds or minutes of the server? by design or bug instead of milliseconds?

Speaking of this, does anyone happen to know around what time they opened Spectre Packs containing N7 weapons? I know Valve did something similar with the Golden Wrench promotion and I've been thinking about it since. I recall getting a Paladin around 7:00 PM EST, but I cannot recall the minute or second.


Also, the whole "buy lower level packs to max out lower level stuff to get better stuff in Spectre packs" strikes me as incorrect. If you max out all Silver weapons and mods, any time you would get a Silver weapon card, it is simply replaced with a Silver Character Card, not a greater chance for a Gold card. Or is that not how it works?

Modifié par Sypher90, 30 mars 2012 - 07:22 .

[Fieryeel]

Sypher90 wrote...

Also, the whole "buy lower level packs to max out lower level stuff to get better stuff in Spectre packs" strikes me as incorrect. If you max out all Silver weapons and mods, any time you would get a Silver weapon card, it is simply replaced with a Silver Character Card, not a greater chance for a Gold card. Or is that not how it works?

That makes sense too.

I 've been getting a hell lot of common character cards after I maxed out my common weapons.

I need to know if common weapons and common weapon mods are mutually exlusive of each other.

[SofieShepard]

Another developer here. I blew off the rng talk as nutters. But then last night after promoting every one I decided to buy 30+ recruit packs to try to bring up the rest of my common weapons. Plus I figured the class cards would serve a purpose for once.

In the span of 6 packs bought one after another, I got the smg mag upgrade from 1 to 5. That's a heck of a coincidence. And like what was said above, once I had most of the common weapons and mods upgraded i got more one use mods and class cards.