Wednesday, July 10, 2013

Everything You Never Wanted to Know: Decryptors

Seems my tutorials are my most requested feature here on the blog. Unfortunately, I have been lacking an opportunity to write something that isn't already well covered elsewhere. Thankfully, Odyssey is an unending treasure trove of industry changes and provides the perfect opportunity to write a shiny new tutorial.

Decryptors - Augmenting Your Invention Results

I glanced over decryptors in my original invention tutorial. I had to skip the finer details and just leave the TL;DR that decryptors are an "advanced topic" for later discussion. The part I left out is that the largest revenue drivers really center around using decryptors effectively. As an operation evolves away from ISK being the limiting factor, and ambition moves toward ship production, it's absolutely critical to understand how decryptors work and how they can augment the outputs for maximum profit.

Before diving into the :math: and specifics, decryptors augment the invention results. Also, it's important to remember that decryptors are completely optional to the invention process.

Several classes of decryptor exist, and each class can modify the following:
  • Chance of Success (P[success])
  • Material Efficiency (ME)
  • Production Efficiency (PE)
  • Result Runs
By burning ONE decryptor along with the datacores in a single invention attempt, the resultant BPC will be different than default, if you succeed. This allows for a wide range of possibilities:
  • Need just one BPC with minimum attempts?
  • Need a large volume of final product?
  • Need the build to go faster than default?
  • Want to spend the least on materials?
Unfortunately, the interface and :math: are even less obvious than regular invention. Allow me to share some of my math and intuition so as to demystify the entire topic.

Decryptor :Math:

Copying

Before breaking down the decryptor accounting, it's important to review how copying works with invention.

For most modules/ammo/drones:
Runs/max_runs=x/10 runs T2 BPC
For ships, rigs, and some modules
 Runs/max_runs=1 run T2 BPC
That is to say, for modules/ammo/drones, it is important to have "full run" BPCs, where with ships/rigs/etc single-run BPCs are preferred.  Since some decryptors change the resultant runs, it's important to know which BPOs need full-run copies and which only need 1-run copies.  This will save you enormous ammounts of time that would outherwise be wasted.

The key to drive home here is: you need to know how many runs the resultant T2 BPC will have BEFORE applying decryptor modifiers.

Order of Operations

The order of operations for applying decryptor augmentations is a little counter intuitive.  The CHANCE modifier applies at the invent step, modifying the chance for success.  Runs, ME, and PE modifiers happen on top of the base result.  This can be a little difficult to understand at first, but once you understand the process, it is pretty easy to apply the modifiers correctly.

NOTE: EVE-id and EVEHQ are WRONG about the runs calculation.  The math was changed in Inferno and bpcs that default to 1-run do not require max-run T1 BPCs to apply extra runs correctly.  10-run BPCs still apply decryptors as expected (except in 1-run case).

Just to drive the point home, using a module case:

Probability Modifiers


The calculation for invention chance is not exactly user friendly. Decryptors add another level of obfuscation to the already weird calculation. Personally, I cheated and just copied the results from EVEHQ’s Prism tool, but some of the more adventurous out there might want to use the datadump for something a little more robust. Unfortunately, the queries aren’t exactly blog friendly, so I have included a link to a more complete walkthrough here.

The TL;DR on probability modifiers is they augment the initial chance for success. For example, if a decryptor had nothing but a probability modifier, you would still get the same default BPC, but instead get them more (or less) often as per the new weight of the coin toss.

Read more about the hardcore-code at the snippet page (UNDER CONSTRUCTION)

Accounting for Decryptors

Much like the :math: for Datacores, Decryptors follow a similar logic. For the purpose of making “good enough” predictions, some assumptions must be made:
  • Variance due to random probability will wash out with volume
  • Both new ME and decryptor use must be accounted for
  • Focus on generating correct price per unit (PPU)
When I account for decryptors the math is as follows:


Lastly, I like to standardize values by “Yield Per BPO Copied”. This gives a decent center point to start to compare options intelligently. You can also aim for “Max Built in a Week” as a compare point, but you may run into issues in actually inventing enough to actually meet those volumes. Also, don’t forget your base case when comparing options. A lot of blueprints will still be viable in their base case.

Example Case


Decrypt All The Things!

Now that we understand the accounting and effects of decryptors, let’s talk about when to use decryptors. First and foremost, it’s important to repeat that decryptors are completely optional and for a lot of products they may be a terrible idea. For instance, most modules will want to steer clear of decryptor use, where ships it’s usually advised to use them always.

As always, check your math before making investments. Though there are some general guidelines on what to use and why. In no particular order.

Also, remember that decryptors are racial specific. Thankfully, the decryptor names now match their relevant Data Interface.
Amarr = Occult
Caldari = Esoteric
Gallente = Incognito
Minmatar = Cryptic

Augmentation

  • +9 Runs
  • -2 ME
  • +1 PE
  • P[modifier]= 60% (-40%)
This decryptor is one of my favorites, but is usually malaligned. This decryptor screams terrible with both a ME penalty and a P[success] penalty. Though, when what you need is high volume on a product that is difficult to produce, this is the clear choice.

Personally, I really like this decryptor for T2 frigates, and certain modules where copying is prohibitive. With the invention time and very large %margins on frigates it can be difficult to really get enough units to market. With the Augmentation decryptor, you trade a lower number of BPC’s for a much higher final number of products. Also, this decryptor tends to be very cheap, and once you factor it against the +9 runs, its cost becomes close to negligible.

Though watch out, the critical metric in T2 production is ME. Each tick lower increases the waste factor by 10%, which means the resultant BPC will be 20% more expensive than the base result. This is a decryptor that is awesome where its needed to augment volume but largely useless otherwise.

Optimized Augmentation

  • +7 Runs
  • +2 ME
  • +0 PE
  • P[modifier] = 90% (-10%)
Odyssey introduced two “optimized” decryptors to pair with the old “best” and “worst” decryptors. This one is very interesting because it pairs the added runs with an added ME. So, where the base Augmentation decryptor was a trade off between volume and price-per-unit, the Optimized Augmentation variant trades that choice for a win on all fronts. Even the -10% P[success] modifier is nearly negligible, when you factor some items base-chances.

But, obvious winner is obvious, and the market is quickly settling into a place where this choice is well balanced against its base counterpart. I personally have taken the opportunity to stock some of these decryptors for a rainy day, as a purpose to quickly pounce on something with a lot of volume when the price spikes (+2 ME lowers PPU), but I haven’t been won over to abandon its base counterpart yet.

Symmetry

  • +2 Runs
  • +1 ME
  • +4 PE
  • P[modifier] = 100% (0%)
This one is a bit of the odd-child in my book. None of the modifiers put it as a clear winner, and it never hits a price to balance against the other choices. In all fronts it’s second-best. And when it is put up against the new Parity decryptor, it looks even more anemic.

It does improve ME, but not as much as Process.
It does improve runs, but not as much as Augmentation
It does improve PE, but not as much as Accelerant

It doesn’t harm P[chance], but Attainment, Accelerant, Process, and Parity improve P[chance]
The only place I’ve found these as a worthwhile choice is with interdictors, since Augmentation yields are a bit too high for the daily volume, and the ME modifier works out nicely... but beyond that I can’t think of many nice things to say about this decryptor.

 

Process

  • +0 Runs
  • +3 ME
  • +3 PE
  • P[modifier] = 110% (+10%)
This is another extremely popular decryptor. With best-in-class ME modifier, a generous PE modifier, and even a boost to P[chance], this wins on all fronts. Since ME is the critical metric, this is generally the decryptor to use for best PPU. It’s extremely popular for cruiser hulls and larger. Also, I have enjoyed a boosted isk/hr using these to churn large T2 ammo at nearly double the rate of base.

Though being popular comes with a price tag, and Process datacores are definitely pricey. I find that a significant cost of my stockpiling effort is spent on keeping Process decryptors stocked and the unbuilt BPC’s they yield as backstock. Also, with no run modifier, these tend to be the most expensive per-attempt to use.

Accelerant

  • +1 Run
  • + 2 ME
  • + 5 PE
  • P[modifier] = 120% (+20%)
Accelerant is pretty simple and lives up to its name: the resultant BPC will build the fastest possible. This is an excellent choice on very long build time items, and I personally favor it for large projects like Jump Freighters (but your mileage may vary).

Unfortunately, accounting for absolute profit is way less effort than optimizing on ISK/hr, so I haven’t seen much use for it personally. Unless you absolutely-positively need it yesterday (tm), I would not condone regular use of this decryptor... but it’s a godsend in the very few cases that its needed.

Attainment

  • +4 Runs
  • -1 ME
  • +2 PE
  • P[modifier] = 180% (+80%)
This is largely considered the “best” decryptor because of its boosts, but I disagree. Yes, you’re nearly guaranteed success, and it comes with a decent PE boost to match the additional runs, but it’s extremely pricey and has a ME penalty. Amateurs are going to look to this decryptor to augment lackluster skills, and anyone looking to “just build one” (which is wrong).

The only possible place you should ever consider this decryptor is where your “cost per attempt” is very high. Jump freighters have a high datacore use (64/try) and BPCs can cost 100M or more. In this case, forgoing a failure can be seen as an added benefit. Unfortunately, +10% on a 5B ISK build is 500M, and you’re stuck with an extra 4 runs than just the “one for DIY”.

If you run the numbers, it becomes pretty clear this decryptor is awful. If you come by it, feel free to make a buck on those that disagree, but even taking 1-2 extra attempts with a Process decryptor would save you more money than saving those runs with a -3 ME/-2 PE blueprint.

Optimized Attainment

  • +2 Runs
  • +1 ME
  • -1 PE
  • P[modifier] = 190% (+90%)
This is a much more interesting decryptor compared with its base counterpart. Pairing a higher P[success] modifier with a boosted ME and less runs is a much better prospect. This still strands the “one for DIY” with two unused runs on a mediocre BPC, but provides a much better set of bonuses.

In my own tools I see Optimized/base Attainment running the same balance as the pair of Augmentation decryptors. In this case, on the very-large build of Jump Freighters I’d pick this before the base attainment decryptor, but I’d still rank Process and Accelerant as better choices personally.

Parity

  • +3 Runs
  • +1 ME
  • -1 PE
  • P[modifier] = 150% (+50%)
Parity is another oddball decryptor, like Symmetry, where I am not sure where it falls for use. Technically, I think it makes Symmetry obsolete, but the mix of bonuses doesn’t fulfil any particular best-use case.

If I have to say something useful about it, I probably would swap my use case for interdictors from Symmetry to Parity. With it being so new, the prices may not be completely correct and it might find a niche. Unfortunately, I can’t think of many places where this decryptor shines as an excellent use case.