Reverse Engineer 20% Broken

I've seen plenty of posts where people tracked their stats of RE's and the larger the sample, the closer to the expected results it was.

A while back I started tracking stats myself after a string of bad luck, and this is what I came up with at the time:

691 greens RE'ed - 132 successful (19%)
316 blues REed - 35 successful (11%)

This was close enough to the 20%/10% tooltip that I stopped counting.

But just now I asked my Magic 8-Ball if you would accept this answer and it replied "Outlook is Bleak"

People are never happy. If RE'ing had a really high chance to occur then everyone would be walking around with the same schematics thus making the auction house flooded with the same items all the time.

By having RE have a low chance to happen this causes several things to happen that are actually good for the game's economy.

1. Takes crafting materials out of the economy, which means people have to farm more to keep up with the demand. As long as there is a demand people will be supplying the materials needed, thus giving both crafters and gatherers something to do.

2. Slows down the rate at which people learn various schematics. This helps control the rate at which items are put on the auction house. If everyone learned how to make the same item at the same time then the auction house becomes flooded, driving down demend, but raising supply, which causes prices to crash as people compete to see their item before someone else does.

Granted sooner or later the person next to me is going to learn the same items that I have, whether it happens tomorrow or next week. But it goes back to him needing to keep buying raw materials or farming them himself, which then drives the economy.

I know many of the people reading this are either going to go against everything I have said, not read this at all, or argue against everything I have said. That's all well and good, this is my opinion I am by no means an expert on the mathematical formulas that Bioware uses to determine RE chance nor do I know exactly how the in-game economy works. I am simply trying to develop and talk about valid points as to why RE'ing has a low chance and why a higher chance would actually harm the economy.

It doesn't have to be a higher chance, though. Instead of a random 20% chance per RE, they could simply require exactly 5 REs to learn the schematic. It's the same overall probability, but it ensures that you can't be unlucky and fail to learn a schem 50+ times in a row.

They could something like the augment kits mats. Everytime you RE something and fail you get like a "token" of some sort (could be go straight to your mission items tab). Once you accumulate like 30 or 40 of them (whatever the "right" numbe should be), the next RE will be a success. This shoul be tiered ofc, so you won't be REing garbage to help you on high level stuff. With something like this, a fail is not a complete fail and you won't be hit by an awful streak of bad luck.

You're doing well so far, but...

Oh, but it is! You're falling victim to something called the Gambler's Fallacy.

If you throw a six-sided die six times, the probability that you will get 1.2.3.4.5.6 or some arrangement of those is exactly the same as the probability that you would get 6.6.6.6.6.6 or any other combination, because each toss of the die is independent of the other. It doesn't matter if you throw the die 5 times, 50 times, 500 times, or 500,000 times -- none of those numbers are large enough for the Law of Large Numbers to come into play. Quick explanation of that is: The measured rate of an event occurring will approach its expected chance of occurring as the number of measured events approaches infinity. Now, it is certainly a reasonable expectation to see things "average out" in a much smaller sample, but for one person at any given time to experience a dry spell of 100 in a row? It's going to be common -- it just happened to me. 1000 in a row? Not quite so common, but not totally ... well, "unreasonable."

The whole idea of your success rate converging on that expected value of 20% or 10% only applies for every single RE event you do plus every single RE event everyone else does. You can't grab some subset of 50 tries and say "That was a string of fails, so I'm bound to get a string of successes" and know that with any certainty. A string of 50 fails would be precisely as likely to occur as a string of 50 successes if it wasn't for one thing: we do not have an equal chance to succeed or fail. We have at best an 80% chance to fail and at worse a 90% chance to fail and, since results are heavily weighted towards failure, it's much more likely to get a long string of failures than it is to get a long string of successes. So, although I recently saw a string of about 50 failures in a row on blue 10% items as well as a recent string of 3 consecutive successes on blue 10% items, I sincerely doubt that the success string will be repeated anytime soon, if ever in the course of this game. A long string of consecutive failures? More likely. Those, as they say, are the odds.

This is all why a purely RNG-driven system is such utter nonsense. Randomness doesn't make things hard or challenging, it makes them futile or serendipitous. There is no skill involved. Would you put up with a pure RNG determination for our combat skills? Yes, combat has randomness to it. There is a chance that you will miss on your attack or heal. But the game gives us ways to mitigate that chance. We can get our mitigation -- our Accuracy -- over 100%, making a hit guaranteed unless the target has some way to reduce our mitigation. Even actual crafting has mitigation of a sort: (1) you have a base chance to crit on a craft, (2) your skill compared to the skill level required affects your chance to crit, (3) your affection score, if maxed, for the companion crafting increases your chance to crit, and (4) your companion may have a skill modifier from +1% to +5%. All of this, if you choose wisely, will get you at best an improvement to a hard cap of 25% chance to crit -- you have your "skill" maxed out, and you still have a 75% chance to fail each and every time. And what does that crit get you? For equipable items, an augment slot. Arguably, this is worthless now that 1.3 has come out since you get a 100% chance of success for an augment slot if you craft a kit and apply it for a couple thousand credits. Given the amount of money you can spend on resources failing to produce an item with an augment slot, going with a kit is the smart money. That new Legacy Perk for crafting? A total of 3% improvement, to in rare situations a cap of 72% chance to fail? Is it really worth 350,000 credits, or is it a waste of money?

Getting back to RE -- what do those factors have to do with RE chance for success? Does your skill level matter? Does the affection level of the companion who crafted the item you RE matter? Does any skill bonus the companion who crafted the item you RE matter? Does the Legacy Perk matter? The answer is a clear, emphatic no. You cannot mitigate your failure rate for RE. Any thoughts of improving your chance to RE by failing a lot, by REing a lot of the same item, by sticking to low-level items to RE ... those are all fantasies. That simply is not how probability works, and so it's not how RE in this game works.

So yes, again, RE is broken. Not because you fail and fail at it and it gets frustrating, but because the system does not allow you, no matter how "skillful" you are, to improve your chance to be successful at REing items.

But how slow is reasonable?

If you know your crafting and REing, you know there are 3 blue prefix schematics to every equipable green schematic you can buy from your trainer, and 14 purple prefix schematics that come from those 3 blues. Some of those purple combinations are pointless -- example, a scattergun with shield chance (yes, shields and scatters are both off-hand and mutually exclusive and yes, the devs were stupid enough to put them in the game). I can imagine the crafters-at-heart out there, like me, think of one day having all those purple schematics all the same for what they can craft. Fact is, you may need to get the useless ones since you currently have absolutely no control over which schematic you get from an RE success ... more mindless RNG crap.

But how long would it take to get all those purple schematics?

I did the calculations for Armstech elsewhere; follow the link if you are interested. My calculations assume that your success rate will approach your chance for success when REing to get new schematics, and this is a fairly reasonable assumption -- for only the equipable items that begin as green schematics for Armstech, it will take you on average around 75,000 items crafted to get all your purple schematics (around 72,000 of them being blue items). That is about 1.5 years of running crew skill missions 24/7/365. Of course, if you get all your companions to max affection before you start crafting and run all 5 companions 24/7, that should cut your time down to a little less than 3 months of crafting. Good luck getting some sleep or something to eat in that time!

Is that reasonable?

It's not the worst of it, though. In order to prepare for that crafting grind, you need crafting mats. I'll leave out the scavenging mats and fluxes, since those are really small investments ... a matter of several weeks of 24/7 harvesting and buying from a vendor (to save time -- you'll need it). For your Investigation materials, it will take something more like 61 years of crew skill missions, running them 24/7/365. Running 5 companions at a time, max affection, increased yield on missions will probably cut that to between 8 and 10 real years, if you are lucky enough to get 5 Research Compound missions in a row for the mats you currently need. And once you finish this, THEN you will have the mats for that 3 month/1.5 crew years grind of actually crafting items. Cost in credits? Expect something in the neighborhood of 15 million credits overall.

What will you have that you can use for yourself, give to friends, sell on the GTN? Nothing. These calculations are simply for what you need to RE items to get schematics. You have your schematics, but you have no goods to show for it.

So, is that slow enough for you, given the system as is?

I don't disagree with the points you made at all.

However, I feel that "broken" isn't the correct term. Broken (in software at least) implies bugged or not working as intended. The RE system is, in fact, working as the developers intended.

It may be a good system, a poor system, or a system that is terrible and needs to be rebuilt. But most of the posts claiming the system is broken are describing a situation of "My results don't match the tooltip percentage"

Yeah, you're right. When I say "broken," I mean that the very idea itself is bad. That's not very precise.

LOL, your Magic-8 ball is working as expected.

I have run three attempts at REing in a row that resulted in 1.) 40+ tries no hit gave up, 2.) 57 tries to hit, and 3.) 15 tries to hit when the tooltip said I had a 20% chance of success. Does that prove anything? No. It indicates that the system may be broken but nothing more. Does your experience prove that the system is not broken? No. It indicates that it may not be and nothing more.

Why should anyone accept your answer over the posts of others that have had different results? My point was that, like your post, the assertions that the system (supported or not by individual experiece) is not broken are no more valid than the those that say it is broken. Neither claim is provable. As has been ponted out so many many times, a players bad experience does not prove that the system is broken. But the converse is also true, a players experience that the system is yeilding the tooltip results, does not prove that the system is not broken. Given enough players and considering the amount of REing being done, one would expect that a broken system could provide a significant number of players with the advertised results.

Broken can also mean a number of different things. A broken system can indeed yeild the expected average results as you have experienced, but have a flaw that produces long steaks. As a lame example, if the chance of success is advertised as 20% but is actually 10% for twelve hours of the day and 30% of twelve hours of the day then if an equal number of tries happen in each twelve hours, the chance averages to 20%. One player (the first twelve hours) thinks that the system is broken and never gets 20%, one player is laughing up his/her sleeve because they are beating the odds, and one who plays during both time segments is getting around 20% and challenges (with his/her numbers) everyone who says that the system is indeed broken and they are not getting the 20%.

^^

I get where you're coming from on that.

From my perspective, one of two things could be happening:

1. Bioware developers wrote a simple 20%/10% rng program that is working correctly. Some people get the expected results, some people get better than the expected results and some people get worse than the expected results. The happy people rarely express their voices on the forums. The unhappy people are worried something is wrong and post.

2. Bioware developers wrote a much more complex rng calculation, or wrote an rng program that is buggy, or they wrote one with a chance lower than the 20%/10% tooltip. People who are refuting the claims of broken and post their results are actually outliers, and even though their results match the tooltip (for the most part) they are, in fact, getting MORE than the program should deliver.

Sure, #2 could possibly be happening -- it's just far-fetched and no where near as likely as #1 above.

Most posts I've seen about RE and percentages are

a) RE doesn't work, here's my small sample size where I show it's broken.
b) RE doesn't work, because <insert gambler's fallacy>
c) RE works, here's my larger sample size where I show it works.
d) RE might work, here's my larger sample size where it's close but still a little low

If more people had posted, "here's my 1000 RE's and I only got 91 successful greens to blues, that's only 9.1% instead of 20%" I would take notice. But I don't see those.

So I say the system is functioning as intended. Doesn't mean it doesn't suck or it's a great system or anything. Just that the tooltip is correct.