Re-evaluating Darwin

[daman_asha]

My last Icebreaker analysis copped a bit of flak on the Adelaide City Grid Facebook group that I run . Some commenters felt that I was unfairly critical of a certain Icebreaker:




Darwin, an evolution in AI ‘breakers?

I stand by my assessment of the card in the context of that essay; the numbers tell a damning story in the isolated assessment that my software produces. I also stand by my assumption of three virus counters and so three strength as a realistic point at which to evaluate the card’s strength. I can back this up with anecdotal support now, having actually decided to play the card. Rarely did I manage to get Darwin to sit much higher than three virus counters before the Corp player decided to purge; admittedly though this decision was not usually taken in isolation of other virus cards loaded and threatening the Corp.

Crucially for the analysis, assuming the “base” strength of three, the application of two The Personal Touch (or equivalently one Dinosaurus, The Helpful AI, or two Datasucker virus counters) does not improve Darwin’s performance against other Icebreakers when considering the average break cost across all the Ice in a particular class (i.e. subtype - Sentry, Barrier or Code Gate).

But now I want to return to Darwin for several reasons. Firstly to address some of the considerations and critiques raised. Secondly, to present the first Netrunner deck list that I’m going to share in this blog, which actually uses Darwin despite its apparent lacklustre capability in comparison to all the other breakers. Lastly I want to talk a bit about the method of evaluating Darwin (or any other card for that matter) in the context of a deck design.
Replies to My Analysis

For those who came in late… My analysis is based on a couple of assumptions - that the breakers can be ranked according to their average cost to break all the Ice in the type that they are keyed to; that the break cost for a piece of Ice which cannot be broken by a specific breaker is set to 100 (to push the average break cost up); and that the strength of variable breakers like Darwin, or the number of advancement counters (and hence subroutines) in the case of the Weyland Ice is arbitrarily set to three. On that basis, Darwin comes up dead last in all three categories, even supported by a couple of Datasucker counters, Personal Touches, or a Dinosaurus.

To which I received the various critiques of my analysis:

• Darwin can break anything;
• The virus counter Strength boost can be paid for with Cyberfeeder and Spinal Modem (a point that I initially missed);
• That in an Anarch deck with Datasuckers & the full [Static?] breaker suite plus Surge, Darwin is perfect for early game pushes before you are fully set up;
• Darwin can force the Corp to purge virus counters to reduce the strength of the breaker; and
• Darwin gives the Runner confidence to hit any Ice.

Reply to the Replies

In brief, addressing each of the points above:

Darwin can break anything. Yes and so can Crypsis, and for far less credits in general, assuming that the approximate cost for adding a virus counter (as a click) is one credit. The difference primarily comes from the cost to break subroutines, at 1¤ for Crypsis and 2¤ per subroutine for Darwin.

In an Anarch deck with Datasuckers and a full breaker suite there is little need of Darwin (or Crypsis) if it sets up the rest of the breakers fast enough. When you’re resorting into the AI breakers in this instance, you want for a backup breaker to be immediately ready to go to be able to apply the pressure this brings, and Darwin isn’t that breaker. It takes time to get into gear, and so is a poor choice for a backup generic breaker.

Darwin can cause the Corp to purge but not only on its own; in my (limited) experience it might be enough to trigger the Corp to purge when Darwin is at the strength required to threaten the bulk of the Corp’s Ice, but more likely the conjunction of multiple loaded viruses is the trigger point for the Corp to purge. I have also observed that Crypsis can equally inspire the purge action, though again it seems to be more an incidental victim than the primary target of the purge.

And as for instilling confidence, if you pardon the rhetoric: really? I’m still quite leery of an early game Darwin at strength 2 running a Corp with one agenda out and four credits, aren’t you?

The upshot is that Darwin isn’t the greatest of breakers, but given all its flaws - resplendent as they are, as in any AI breaker in this game - it can be worked around. Alone in consideration against any or all of the other breakers, Darwin comes up dead last, but the analysis here is an idealised one: one where I can have any breaker installed for free, where I have infinite MU, and any breaker I like in any number I like in a deck (or just one of each I like which comes up in the opening hand every time ;-) As a backup breaker I think Darwin is weak, itself requiring more support than the main deck it is meant to be supporting can carry. That said, as a primary (or sole) breaker, however, Darwin is perhaps a better choice than any other generic breaker in the game.
The Deck Design Process

I decided that I wanted to try Darwin in a deck, to see if it was possible to make it work. The basic idea was to include enough support cards to overcome the obvious weaknesses, without having to stuff the deck so full of cards to compensate for Darwin that this was all it did; there needed to be space for some cards that would actually advance my strategy and win condition beyond simply running with Darwin.

In addition to wanting to actually give Darwin a ‘fair go’, I also wanted to try out Data Leak Reversal and Retrieval Run. Finally I thought that with the low number of virus cards that I would be using, that this might actually be a deck that worked better for the Whizzard, Master Gamer identity than for the ubiquitous Anarch Noise, Hacker Extraordinaire.

Some inspiration for card selection was drawn from Jose “Hraklea” Almeida’s Root Cause columns on CardGameDB, Economy, Parts 1 & 2, in particular the economic boost that is possible with Joshua B. and Liberated Account, turning a single turn into +10 credits at the cost of taking a tag at the end of the turn, which is as good as Magnum Opus, without the MU cost. While the deck would run Plascrete Carapace as Scorched Earth defence, the idea was to try the credit denial approach, with the aim of sending the Corp broke such that the threat of Scorch became economically unviable. For Retrieval Run to work I wanted high card draw to overfill my hand, and big expensive (but efficient) Icebreakers like Morning Star and Femme Fatale to maximise the install cost savings that RR gives.

The initial deck design didn’t work; the ‘clunk factor’ was too high. I realised that it was trying to do too many things at once; run and support Darwin, play “tagMe” with the Corp, mill with the DLR, and set up using RR to run late game as efficiently as I could. Time to strip it down, work out what I wanted it to do, and focus on doing that. This is one of the first deck design heuristics that I would present:

• Heuristic #1: beware trying to do too much in the one deck

I decided that the RR idea could wait for another day, and another Whizzard design (or more likely an Exile, Streethawk design after Creation and Control comes out). Out with the RR, MS, Force of Nature, Mimic and FF, and keeping only Darwin and Crypsis (as a backup breaker for the times when the Corp purges Darwin down to strength zero again). I did want to use the DLR “mill” combo and was happy to stay with the basic tagMe concept. But I also wanted to stick with experimenting with Darwin to see what it could do.

Inspiration was sought and I was lucky to find that another Runner out there was also working on a similar design. I found the Whizzard DLR deck listing by “Solomir” on the CGDB site, and was inspired by the use of Darwin’s Best Friend:




Cyberfeeder, an engine unto itself

What’s so good about Cyberfeeder for Darwin? The Cyberfeeder supports the use of Darwin in three modes:

1. because Darwin is an Icebreaker - AI - Virus, Cyberfeeder can (help) pay for the installation;
2. because the strength function of Darwin is an ability on an Icebreaker, then by extension of the ruling in FAQ 1.1 regarding the interaction between Femme Fatale and Cyberfeeder (i.e. being able to pay for the bypass ability using Cyberfeeder recurring credits, as this is considered using an Icebreaker) Cyberfeeder can pay for the addition of virus counters to Darwin; and
3. the usual mechanism of Cyberfeeder’s recurring credits in paying for the raise strength / break subroutine abilities on Icebreakers.

After that the decisions came moderately easily; I swapped out Grimoire for Spinal Modem, despite the risky nature of the latter console in the current trace heavy environment. The extra credits for breaking mean more to me than the free virus counter, although that can certainly accelerate the deck. The downside is that to reap the benefit of the Grimoire, I need to have it installed before the Virus cards come into play, whereas the credit acceleration from Spinal Modem comes whenever it arrives, independent of the order of installation. Playing Dinosaurus-oriented Shaper decks has taught me that holding an Icebreaker that is reliant on having the Dino down isn't a great plan, so I am leery of being stuck in a no-console limbo. The addition of E3 Feedback Chip also supports Darwin, by making the break cost 3¤/2 subroutines instead of 4¤, or 4¤/3 subroutines instead of 6¤. E3s are also (obviously) good for attacking Haas-Bioroid's Bioroid subtype Ice when you have no breakers installed and have to click-break them.
More about Darwin - Reaching for Synergy

Darwin, like any other AI Icebreaker, needs support, which is one of the reasons that it performs so badly in the in-isolation Icebreaker analysis that I conduct. The potential advantage of finding support cards for a breaker like Darwin is the potential for them to synergise with the rest of the deck. Cyberfeeder is a great example of this; not only does it support Darwin in three modes, but it can help to pay for installing other cards that support Darwin in other ways, such as Datasucker and Parasite.

Parasite works to passively speed up the deck by targeting big Ice (high strength, or perhaps worse for Darwin, multi-subroutine Ice) so that each turn I'm putting a 4+ strength Ice closer to being broken by Darwin by two - one for Darwin's "buff" and one for Parasite's weakening of the Ice. Datasucker requires more effort, but one or more of these viruses can help to pull out-of-range Ice down to breakable strength, or can even synergise with Parasite to drop the Ice strength to zero, eliminating the need to break the Ice altogether by destroying it. Perhaps the strength of Crypsis-only (or Crypsis-primary) decks comes not so much from the generic break functionality of Crypsis, but the power of that breaker, plus the aforementioned Viruses working in conjunction with one another?

Returning to the design process, Cyberfeeder's economic advantage leaves me free to use my 'real' credits for other things, meeting crucial credit generation thresholds such as 5¤ for Sure Gamble, or 6¤ for Liberated Account. Spinal Modem, as mentioned, is a dangerous option in the trace-heavy environment of post-Genesis, but the risk/reward seems worth it, and the two recurring credits seem a natural fit for Darwin's break cost. The Spinal Modem can also play two of the roles of the Cyberfeeders in support of Darwin, but its credits do not support Virus installs.

I also chose to run with Djinn, which not only fetches any of the missing Virus cards from my Rig (Darwin, Datasucker, Parasite, ...) but also provides storage space for the Datasuckers if not the Parasites, and the sole Nerve Agent if I need to switch to dedicated HQ running. I run Crypsis as a backup breaker for instances when the Corp chooses to purge the virus counters. This makes for a formidable attack array capable of putting severe pressure on the Corp, especially as the Parasites start to take hold and destroy their host Ice, at which point Déjà Vu can retrieve them for ongoing pressure.
Evaluating Cards: Combo, Synergy, Force Multiplication and EBA

This brings me to some of the observations that I want to make about evaluating cards, in isolation versus within the context of a deck design. One mistake that many pundits and commentators repeatedly make is to consider cards only in isolation. While there is some advantage in doing this, as each card needs to be evaluated on its own merits when considering whether it is a good card (overall or for inclusion in a specific deck), this fails to address the whole picture. For one thing, it is impossible to truly consider whether a card is a good card ex nihilo, because the evaluation must be done in the context of the entire card pool. Ignoring the rest of the environment makes as much sense as evaluating a blank piece of cardboard and deciding that it is good or bad; the question is whether it is good or bad in relation to what?

When evaluating cards in the context of a whole environment there are many ways to consider its value or function. You can look for combo's or synergy, and you might to consider force multiplication and an effects-based analysis. I want to define each of these a little more and give some examples before returning to the deck and presenting the list as it stands.

“Combo”, as we may well all know, derives from “combination” and the term has been around since the early days of Magic: the Gathering. Rather than reach for other’s definitions, I will use my own: “the use of one card in conjunction with another that provides a greater effect than either in isolation.” NB: there is no constraint on the notion of a combo to be restricted to only two cards; it is possible to have triple-combos (e.g. Cell Portal + Neural Katana + Whirlpool).

I want to distinguish this from “synergy”, however, because I believe synergy in reference to customisable card games has a broader scope than combos. Perhaps I should define synergy to be “the interaction between the cards in a deck, or a subset of a deck Σ, (where |Σ| > 2, and preferably >> 2) which improves the functionality, effectiveness or utility of the majority of cards in Σ.

To give some concrete examples: Darwin + Spinal Modem is a combo, as is Parasite + Surge. Darwin, Datasucker, Parasite, Spinal Modem, and Surge together have synergy. While it may make sense to say that two cards have synergy with one another, this is really only identifying a combo. The distinguishing feature between combos and synergy may be that a combo has one instance of use and that one of the pair (triple, etc.) of cards may be useless or ineffectual on its own (e.g. Surge without a Virus). Synergy is like aggregating a collection of combos, allowing individual combinations of the synergy set (Σ) to function effectively even in the absence of other members of the set. Darwin and Datasuckers, without Cyberfeeder, still have synergy, do comprise a combo, and can be effective.

• Heuristic #2: seek out synergy over combos in your deck design

Force Multiplication (FM) is another factor to consider in contemplating and analysing cards. I’ll define FM as “the effect of a card being escalated by the presence of multiple copies in play.” Another example is called for: aside from the obvious advantage of having three copies of R&D Interface in a 45 card deck lifting the odds of any one draw being that card from 1/45 to 1/15, there is a FM effect. Most people, in contemplating R&D Interface for inclusion in a deck consider the advantage of having more than one in play at a time, allowing more of R&D to be accessed over less runs; more effect for less effort being expended.

As another example of this concept, when Compromised Employee was released in Trace Amount it was pretty much panned by the CGDB “staff review” panel. The initial assessment was primarily economic and I’ll paraphrase it as pretty much ‘Compromised Employee costs 2 to install, so the Corp will have to rez three pieces of Ice before CE will return a profit.’ Setting aside the notion that a Corp deck not rezzing Ice to defend itself being a very strange deck indeed, the assessment neglects to consider the effect of multiple CEs in play. It also doesn’t really take into consideration that the cost to the Corp to rez the Ice is likely to be higher than the cost to install one or two Compromised Employees, or that the CE can ‘pay’ for itself in other ways by providing link value.

Granted, each installed CE needs to recoup the install costs itself (and so looks like this is putting the Runner further into the hole)  but with multiple installed CEs, each instance of a rez now triggers two or more effects. In situations like this, as the facing Corp I have found myself reticent to rez Ice because of the credits that it would give the Runner. The credits are also gained mid-run, which can contribute to the effectiveness of the run, and may be the difference between success or failure.

The short form of FM is “if one of the card is good, then two of the card is better.” Not all cards do have a force multiplier effect, but it is inarguably a factor in the design of cards in the game. Consider, for example, the impact that Ice Carver (or Xanadu) could have on the game if it were not Unique. Joshua B. (setting aside his Uniqueness) and Deep Thought are good exemplars of cards that do not force multiply. Medium, R&D Interface, Underworld Contact and Datasucker are good exemplars of cards that do.

• Heuristic #3: force multiplication reduces the work effort required for effects

Lastly I want to consider what I will call Effects-based Analysis (EBA). By this I mean “consideration of the effect of a card (or set of cards) on the overall game state.” While this may be related to synergy in some cases by having the effect of making one or more cards more efficient or powerful (and so magnifying the effect that they have on the game state), it can also mean a more generalised impact on the game state and the advantage that this can bring (in credits, clicks, or what other resources might be contested).

EBA looks to consider how certain cards or combinations of cards can impact on the game and advance the strategy of the deck design. if the concept behind a Runner deck is to, for example, wrest an economic advantage over the Corp, then CE in multiples can help to do that. As the Corp goes down in credits by paying to rez Ice, the Runner gains credits. Combining this with an HQ run and playing Emergency Shutdown, the imbalance created is increased as the Corp loses even the advantage of having a rezzed piece of Ice. So the effect becomes greater still in the consideration of the EBA; the economic advantage is being steadily turned over to the Runner. Under an EBA, the potent card of Emergency Shutdown becomes even stronger, as a second run at the targeted server forces more credit expenditure from the Corp and more gain by the Runner.

As an example related to the deck in question (which I have not fully tested), prior to the last tournament I had a provisional list at 43 cards. Being in the enviable position of having two slots (and swathes of Influence) remaining, I was able to consider two-ofs such as:

1. Surge;
2. Diesel; or
3. Mr Li;
4. Ice Carver;
5. Xanadu;
6. The Personal Touch;
7. Personal Workshop; or even
8. Repicator.

While the final choice ahead of the tournament was to go with two Surges for the synergy that they have with the program suite, the decision making leading up to this is perhaps more interesting. I chose to discount Mr Li because of his poor synergy with my other draw mechanism, Quality Time. This is my O:NR bias showing here, as the two cards’ precursors: Crash Everett, Inventive Fixer and Bodyweight Synthetic Blood, did have a good synergy. The lack of ability to combine Mr Li’s action to double draw with the action to play Quality Time put me off the inclusion. A very viable choice, Ice Carver, would be suggested for the synergy that it can bring to interactions with Darwin and Parasite. Finally EBA would suggest Xanadu was the appropriate choice, making it cost more for the Corp to rez Ice which hopefully I would be able to destroy on subsequent terms the using Parasite. Subsequent to a properly set-up Vamp, even Ice Wall becomes a bridge too far for the Corp to rez, allowing me freer access to remotes/R&D/Archives and thus cheaper running.

The decision not to do this comes about because of a final consideration in evaluating a card for inclusion in a deck - metagame. Much (very much!) has been written about this notion (here, here and here for a Wikipedia link, or even here at tvtropes for some quick examples), and so I won't set about rehashing this well-worn topic in this essay. Suffice to say, my local meta is such that Foxfire does see some play, enough that Virtual resources are vulnerable, and I would rather not waste deck slots as well as the install actions and three credits on a resource that the Corp can waste for an action and card on the next turn.

This brings me to a couple more heuristics for card evaluation:

• Heuristics #4: consider cards in light of EBA as much as synergy and FM
• #5: do not discount the local metagame when determining card choices
• #6: when evaluating cards, aim for synergy, and consider FM and EBA; if a certain card meets all three criteria, there's a good chance that it is already in the pile of "must run" cards you have pre-selected;
• #7: question assumptions and given wisdoms - you may well discover a novel application for a card that nobody (at least in the local game environment) has yet hit upon; doing the unexpected can win games.

The Deck - Whizzard: “Dumpster Diving”

I should finally present the deck list and finish off the essay by discussing the remaining card choices. The name comes from the original design and was meant to reflect that Whizzard was rifling through his own heap for programs as much as the Corp’s archives for Agenda.




On reflection and from play, I found that the second Crypsis was redundant and I would swap that for a third Djinn. The primary awkwardness is trying to get the combination of Joshua B., Vamp and Data Leak Reversal to all come together at times, but the deck can certainly pressure to the win without it. Playing Joshua B., and DLR without the Vamp leaves the components vulnerable to the Corp’s trash action, but can be enough to draw the into doing so and ignoring the defences, which might be enough.

I found myself initially missing the Magnum Opus from the original incarnation until I re-learned the truth of Anarch: run poor if you have to. I have won with this deck with a very sparse rig and almost no money. That said there should be plenty of money to be had along the way with  the Liberated Accounts and Kati Jones. She is vulnerable to the Corp destruction effect once tagged, but “emptying the broker” before you start accumulating tags works well, and once you’re at the point of no longer caring about tags, if the Corp wastes an action to remove her - even when you have invested an action of your own to load her, you’re up by an action in the late stage of the game, or two if Joshua B. is still around.

There is a definite point in the game where tags no longer matter, and I’ve just started hammering Joshua B. every turn. With the Corp under sufficient economic pressure and staring down the barrel of a well primed Darwin, building Parasites and Datasuckers, it can be a hard choice for them to either purge or destroy resources. Quality Time can be a tricky card to play, and is something of a legacy card from the earlier version of the deck where I wanted to draw as many cards as possible so as to be able to discard for later Retrieval Runs. It could be revisited to look for alternative draw enhancement. The trade off between credit cost and Influence cost for QT versus Diesel can be a tough call, but in this deck there’s Influence to spare and Diesel is easier to manage.

Most of the other cards I have already discussed in the body of the essay, so there isn’t much more to be said about the decisions underlying their inclusion. I have tweaked the deck for further exploration, experimenting with adding one Account Siphon (which worked very well), trying the aforementioned swap of one Crypsis for a Djinn (good), swapping the QTs for Mr Li (very experimental and I’m still unconvinced here). On the whole the deck is running at (stunningly) around an 89% win rate under tournament and league conditions, albeit from a fairly small sample set. If the deck continues to perform this well, it appears to be an extraordinarily strong Runner.
Darwin: In the Final Analysis

So now that I have played Darwin in a winning deck, what do I think of it as a breaker? Undeniably it still has lots of weak points. On its own Darwin is ineffectual and slow. It may be a decent early game breaker to install for initial runs, potentially busting through nigh-ubiquitous Ice Walls and saving itself from an early Rototurrent (but probably not getting through) or avoiding the tag effect of a Shadow in Trick decks. Against an early rez of Viper, Eli 1.0 (who sees a lot of play at the moment) or even a Wall of Static, Darwin just stops. Corroder and/or Yog.0/ZU.13 Key Master are far better bets for early aggression.

As a backup breaker (instead of Crypsis) I believe it is also weak. If unsupported, Darwin is weak however a supported Darwin is much better; within a deck like the one above there is much more opportunity to pressure the Corp, and to run aggressively.

Is Darwin then a bad card? That’s harder to say, one way or another. I think in the right deck, Darwin can be a powerhouse, serving multiple roles: threatening the Corp’s centrals much earlier than a full Rig is capable of, attracting a purge (though again this seems to be incidental to the purge against Parasite, Datasucker and possibly Nerve Agent threats), and freeing up deck space for credit generation and win condition cards. Used poorly, it isn’t up to snuff, and the fixed strength breakers, or Crypsis as the sole breaker (possibly using Corroder for Barriers) can serve better. Darwin is a mid-range card that can be awesome in a deck designed to work with it, which is an interesting thing to find in any card game.