one million words's picture
By: one million words, Pete Jahn
Dec 13 2019 1:00pm
Login or register to post comments

 State of the MTG Arena Program

As some of you may remember, I used to write SotP, about MTGO. This week I’m writing about Arena. I know a lot about the history of Magic, of MTGO and of Arena, much of which may be relevant. 
For those of you who don’t remember me, I wrote weekly columns on PureMTGO and various other websites for over 20 years. I covered MTGO from the time it was first in closed beta until last April, when I retired from writing a weekly column. I have decades of experience with both paper and digital Magic.
I may have stopped writing weekly columns, but I have not stopped playing Magic. For the last year, I have been playing paper Magic, some MTGO, and a fair amount of Arena. For this article, I am going to grade Arena on a number of criteria and talk about how it compares to MTGO. I’ve been playing MTGO long enough to remember events like Chuck’s Virtual Party, the v2 to v3 conversion, 3D avatars and more. Some of those events – and the problems leading up to or resulting from them – seem eerily similar to what we are seeing now.  
My Arena bona fides: I started playing Arena when Wizards lifted the non-disclosure rules during the beta, and have played pretty much continuously since. I am almost exclusively a free-to-play player.  I bought the $5 introductory package, but I have not spent any money on Arena since then. I generally finish my quests every week, but I don’t always finish my dailies. I play some limited, but mainly play Standard. Given my limited time and cardpool, I generally finish mid to high Silver on both ladders. I watch a number of streamers playing Arena, as well as events like Twitch Rivals, so I see Arena play at all levels.   
I am going to rate the program on several different criteria. What I am not going to include in the rating, though, is the state of Standard. I started writing this before the Field of the Dead ban, much less the Oko and friends bans. Pre-bans, Standard was not all that fun to play. Hopefully, that will change, but the state of Standard is a broader topic. This article focuses on the state of the Arena client.  
Let’s start with one of Arena’s good points:
Rules Engine: A
Like MTGO, Arena uses actual Magic rules, and does it well. It is not a form of Magic Lite, like Duel Masters or Magic for cell phones. Arena uses all the phases and steps, and the cards do what they are intended to do. It works well. The minus, however, is that the client has only a tiny fraction of the entirety of Magic cards currently programmed into the rules engine, and only those that have been created since R&D began designing with digital play in mind. It works well, for now – just like MTGO did back when it was called MODO. (Later on, when the MTGO client had to incorporate cards like Eon Hub, and key words like Phasing and Banding, the MTGO client tended to bog down.) So I give Arena an A for rules engine, for now. I hope the client is robust enough to keep running as well even when we have the entire Pioneer cardpool on the client.      
Availability: B-
Generally Arena is available for play 24-7. This may not seem like much of an accomplishment, unless you lived through the ups and downs – sometimes lengthy downs – of MTGO back in the day. Now we just download updates and keep playing. That’s an improvement for an online Magic game, but it’s standard fare for the other games I play. Still, I’ll consider giving Arena an A for client availability.
What brings the availability grade down is that the only formats that are consistently available are Standard and draft of the current set. The client supports Pauper, Brawl, Singleton, sealed deck and drafting other sets, but those formats are only available on rare occasions. This is fine when Standard is a fun, viable format. It is more of a problem when it isn’t.
Having multiple formats splits the player base and makes it harder to find an opponent.  I remember sitting in queues for long periods, even hours, hoping that someone else wants to play.  I know that Wizards is afraid of having players abandon the client if they have to wait too long to be matched. However, I think Wizards is working too hard on limiting options.
Matchmaking: D+
Wizards had promised us the ability to find opponents playing the same format, of a similar skill level, using comparable decks, rapidly. Of these, Wizards fully delivers on the first: they find me opponents playing the same format. (e.g. I have never been matched against a Standard player when I am playing an M20 draft deck.) Wizard does okay on the last criteria as well: they match us relatively quickly. I have very rarely had to wait more than 60 seconds or so, and generally not more than 30 seconds. For skill level matching – well, I have no idea how Wizards determines skill levels – presumably by bronze-silver-gold-etc. rankings – so I cannot tell how well that works. For the final criteria – matching against similar decks – I have to fail Wizards completely.
As a free-to-play player, I face some serious problems building decks.   I simply do not have the cards.   I have earned enough wildcards in nine months of play to build one and a half Tier One decks.   Standard decks take a ton of Mythics and rares. The Top 8 of the last Mythic Championship had an average of 10.75 Mythics, 30.5 rares, 16.5 uncommons and 3.75 commons per deck. Even the two color decks had a dozen rares in the mana base alone. As a free-to-play player, I cannot create such decks. Neither can new players. To keep those players playing, they need to be able to play against other players with comparable decks. That’s super-important, and that’s where matchmaking comes in.
I play a fair amount of both ranked and unranked best of one Standard. My best decks are mono-colored (because I cannot afford multi-colored mana bases) with a mix of the rares I have opened and those few I have burned wildcards on. They are good, not great. Call them tier 2.5 at beast. However, I kept getting paired against Tier One decks that appeared to be full copies of the 11 Mythic, 30 rare decks I watched on coverage. Since this could have just been confirmation bias (just remembering the events that match my expectations), I decided to do a serious test – to play some decks that should not, if matchmaking actually considers the cards being played, be paired against Tier One decks. 
I tried playing the intro decks, exactly as new player would receive them. I played a number of matches. Most of my opponents had plenty of rares and Mythics, and I saw plenty of netdecks. I think I faced one match against another new player deck, but just one. However, the new player decks have a handful of rares, so that was not a complete test.  
I decided to go one further: I built mono-colored Pauper decks: no rares, and (almost) no uncommons. The decks had decent curves and used what brewing skills I have, but these were pretty much all common decks. (The green one was 100% commons. The red deck had a playset of Chandra’s Spitfires, because I needed something to fill the last slot.)  If the pairing algorithm actually considers deck contents at all, these should be paired way down. 
Here are the lists:
Pauper Red

4 Chandra's Spitfire
4 Embereth Paladin
4 Gingerbrute
4 Rimrock Knight
2 Rubblebelt Recluse
4 Scorch Spitter
4 Spikewheel Acrobat
2 Scorching Dragonfire
1 Searing Barrage
2 Shock
4 Skewer the Critics
1 Maniacal Rage
2 Dwarven Mine
22 Mountain

Just to make sure I was not building garbage, I tried both decks against the Arena bot.  (Yes, the bot is bad, but so are the decks…)  I went 5-0 with both decks. Good enough. It was time to battle, but I waited until the beginning of the month, so my limited and constructed ranks were both bottom of the bronze. I had not played much the prior month, so previous months rankings had been nothing special – just high bronze / low silver. On every criterion I could think of using, it was a reasonable test of Wizards’ statement that matchmaking takes into account the cards you are playing.

I should have been paired against other players with starting level decks. That is not what actually happened. Here are the cards I saw, in the order they were played by my opponents.
Pauper Green in the Standard Ranked best of One queue – starting bottom of bronze


·       Shocklands, Field of the Dead, Teferi, Time Wipe, Beanstalk Giant, Realm-Cloaked Giant
·       Cavalcade of Calamity, Fervent Champion, Torbain
·       Stormfist Crusader, Fervent Champion, shocklands, Midnight Rider
·       Shocklands, Teferi, Fires of Invention, Narset, Sarkhan the Masterless, Kenrith
·       Gilded Goose, Temple of Mastery, Oko, Wicked Wolf, Paradise Druid, Hydroid Krasis, Questing Beast
·       Shocklands, Gilded Goose, Teferi, Oko, Cavalier of Thorns, Nissa, Hydroid Krasis, Wicked Wolf
·       Rankle, shocklands, Liliana, Dreadhoard General, Legion’s End
·       Gilded Goose, Bonecrusher Giant, shocklands, Voracious Hydra, Shifting Ceratops, Once Upon a Time
·       Pelt Collector, Castle Embreth, Bonecrusher Giant, Rosethorn Halberd (not a rare!), Robber of the Rich, Domri, Sarkan Hellkite, Questing Beast, Shock for the win.
·       Scorch Spitter, Footlight Fiend, Slaying Fire, Light up the Stage, Chandra’s Spitfire
0-10, opponents played decks composed primarily of rares and mythics every match, until I hit the Calamities deck. So I was still bottom of bronze, and kinda sick of losing. Still, I had more testing to do. 
Pauper Red in Rankled Standard Best of One w/ 4 Chandra’s Spitfires
·       Mono red elementals deck – I won turn 5
·       Temple of Epiphany, Improbable Alliance, Redcap Melee, Thrill of Possibilities, Merchant of the Vale, Beacon Bolt, Arclight Phoenix, I won with Embereth Paladin
·       (opponent took forever to mulligan, then conceded to my Mountain, Gingerbrute)
·       Scorch Spitter, Light Up the Stage, Runaway Steamkin- typical calamities deck), lost
·       Orzhof Enforcers, Priest of Forgotten Gods, Midnight Rider, Mayhem Devils, Dreadhorde Butcher
I stopped taking notes here – I was getting to annoyed – but I did play out the string. I went 2-8 with the red deck. The last five decks were all rare-heavy net decks. I also played both decks in the unranked queues, with pretty much identical results. 
Conclusion: Matchmaking is quick, works fine for the upper end of the ranked queues, but is massively unfriendly to new players or those with limited collections. This has also been a problem with MTGO at times, but MTGO at least has featured casual rooms or queues. Arena does not. My series of games with these cards – winning just 3 of 45 matches, not counting games against the bot – was not inspiring. Had I been a new player, I can’t imagine wanting to continue.    
Overall Appearance and Performance: B -
Generally, the client looks pretty good. It has lots of splashy animations, which I understand the kids like these days. I can’t say they do anything for me, but apparently I’m not typical. I will admit Arena looks a lot better than MODO did, and that MTGO does. 
Wizards also seems to be making a fair amount of money with special card styles and the pets. That’s fine – the program needs to be profitable.  I bought the fox – partly because I love dogs, and partly because that I’m susceptible to that sort of do-something-every-day-get-bonuses stuff.
Based on looks and glitz, the client gets an A.
On the flip side, Wizards is loading a whole bunch of animations and splashy extras onto the client. All of that takes processing power. Doing this can cause the client to slow down or get buggy. We see some evidence that this is already happening. Streamers are reporting problems streaming the client, even on powerful computers built to run high end games. Players are rebooting the client every couple hours. Worse yet, Wizards is already talking about working on “performance … and overall client health,” and fixing issues related to game crashes and “Gameplay Hitches, FPS Drops, Unresponsiveness.”  (See, for instance, October State of Arena.) That is not a good sign. Long time MTGO players will remember years of those sort of statements, often leading up to serious problems and replacing the entire MTGO client with a new version. We are only a couple months out of beta, and we are already seeing these problems. That is extremely concerning, given that the client still lacks several vital features (discussed below), all of which will put additional strain on the client.  
Draft Bots: C-
Arena has only one form of drafting – drafting against an AI. The AI is programmed to draft “like” a real person – but it is not a real person. It is a program that is based mainly on a preset pick order, coupled with some knowledge of what colors it has drafted and a slight preference for certain colors or archetypes to give each seat some “personality.” It appears to have a smidgeon of randomness so it is not completely predictable. However, the AI is not a real person, and it does not self-correct when drafts go wrong. More importantly, when the AI undervalues certain cards, they can reliably be left to wheel. 
This causes problems. In each format, certain cards are undervalued, and certain strategies become overpowered. For example, in M20, the bots did not take Heart-Piercer Bow, Faerie Miscreant or (Renowned Weaponsmith) at all highly, so it was fairly easy to end up with a deck full of bows and small fliers – a combo that could lock opponents out of the game. In Throne of Eldraine drafts, the AI undervalues the mill decks, so you can routinely draft very strong mill decks. (Late edit note: the Eldraine AI appears to have been fixed. My most recent draft was much better. I only faced one mill deck.)
Since each person is drafting against an AI, all players can end up drafting the same overpowered archetype. It is, at times,  not all that unusual to play seven matches in an Arena draft against the same archetype.  This would never happen in a live draft.
According to Ryan Spain, who designed the AI for Wizards, AI drafts were intended to be an entry into drafting, not a replacement for drafting against real people. The AI drafts are low-pressure, since they are not timed. You can spend as much time as you want on a pick – you can even leave off in the middle of a draft and return a day – or week – later to make your next pick.  This convenience is great for casual players. Me, too. My dogs have needed to go out, or gotten into trouble, several times during online drafts. On Arena, I can take them outside – or stop them from ripping stuff up – without missing picks. But the downside is that the drafts are never as good as when you draft against real people, either on MTGO or in person. I think it says a lot that most of the major streamers now draft on MTGO, except when contractually obligated to play on Arena. (On the plus side, having people draft on MTGO is getting a lot more cards into the MTGO card pool.)
A personal peeve: the bots are also programmed not to pass rares. This does mimic real life, sort of, but this was true even back when the shocklands were being drafted, and even in the second or third packs. In serious Ravnica block drafts, where people are playing to win, they don’t first pick off-color shocklands in pack three - but the bots apparently did. At least, I never saw any pass shocklands in pack three. As someone who drafted a fair amount of Ravnica block, and had to craft nearly all my shocklands, that was disappointing. Since I also draft as a way of accumulating cards for constructed, having bots take rares just to take rares does make me less likely to draft.
Economy:  B 
The Arena economy is totally unlike the paper or MTGO economies. On Arena, cards are not tradeable, nor can you purchase them directly from vendors. On Arena, you either have to open cards in booster packs or craft them from “wildcards.” One plus – when you open or draft a fifth copy of any given card, that card will automatically be converted into wildcards or gems. The main point, however, is that once you obtain cards, they are yours forever. You cannot sell, trade, dust or otherwise convert cards you no longer need into cards you want. This means that, once you commit to a deck, you are pretty much committed to the deck – unless, of course, you spend real money. You can buy gems with a credit card, and convert those gems into packs. Opening those packs will give you cards and wildcards. 
As a free-to-play player, I am not buying tons of packs, so I only accumulate cards by winning events, like drafts, and by completing quests and dailies. This does earn me cards and wildcards. About 6 months ago, I cashed in all my wildcards to build a mon-green deck that was Tier 2, but something I could play in ranked queues. Much of that deck rotated. I have cobbled together a new version just by adding cards I drafted or got as rewards. Now, six months later, I had accumulated enough wildcards that I bought the cards for a decent Gruul deck. I’m committed to green, but might be able to afford a GB control deck. Maybe. 
On the downside, as Wizards keeps changing the format through bannings and new card releases, my deck may no longer be competitive. For example, I built, and almost sank wildcards into, a GW Adventure deck. I’m glad I didn’t. That deck revolved around the Innkeeper, and requires (Once Upon a Time) to find the Innkeeper. With Once Upon a Time banned, the deck was no longer competitive.   That’s a problem, since there is no way to convert a deck into anything else – no trading, no marketplace, and no way of turning existing cards into gold, or gems, or wildcards.     
For someone who really likes a certain archetype and intends to keep playing that archetype, the Arena economy probably works well. For someone who, like me, loves to keep changing decks – and trades away cards from old decks to make new decks – it does not. 
Completeness: C-
What Arena does – let you play Magic digitally – it does fairly well. However, the program is missing some important elements. Really important elements. Here are a couple.
Friends list: Simply put, there isn’t one. Months ago, Wizards polled Arena users on what they wanted in Arena. The number one response was a way to find their friends online – a friends list. MTGO has had one pretty much forever. Arena still doesn’t have one. Wizards says we’ll get one “soon.” The earlier version apparently worked but did not scale. .

(Manager's Note:  The article was submitted before the friends list was added to Arena.  I am keeping this here for completeness)
Replays: So far as I can tell, the only ways to watch replays of your games is to use video recording software to record your games, or play a streamer who posts their own replays. The program does not support replays. We MTGO players have had, and lost, replays any number of times – but the intention was always to make replays a fundamental part of the program.  Not so for Arena.
Tournaments: So far, the only “organized” play on Arena is the equivalent of leagues on MTGO: you enter a format queue and are paired against the first appropriate player. Arena does not support larger events, Swiss pairings or the like. Large Arena events, like Twitch Rivals and the MCs, use outside software to run the event. The event organizers tell players who they are playing and have those players direct-challenge their opponents. That is not an optimal solution.
MTGO has had problems incorporating and handling Swiss tournaments in the past. Several times over the years, Wizards took MTGO tournaments offline for fixes. However, structured tournaments have been a working feature of MTGO for almost all of the decade and a half that MTGO has been around. It still is: it’s how MTGO MCQs and Challenges are run. Arena doesn’t have it, and I haven’t seen any indication it might get there anytime soon.   
Security: Earlier this month, I received an email from Wizards telling me that I had to change my MTGO and Arena passwords. The problem was that Wizards had a file with passwords and account information insufficiently protected, and it got hacked. That’s a serious error, but it is just symptomatic of a deeper problem. Wizards security is not as strong as the other games I play. For me, the most obvious flaw is the lack of two factor authentication. Now I’m not saying that Wizards’ security is bad, but it certainly could be better.
Conclusions: C
What Arena does, it does well. The rules engine works. The game supports Magic the Gathering – at least the Standard portion of the game. It matches you with opponents quickly. It looks far flashier than MTGO, and it has a lot of glitz. Wizards is making money selling pets, and e-sleeves, and special card treatments.  That’s fine. The program needs to make money. 
However, the program seems to be laboring, at least a bit, under the processing load it already has. Wizards is talking about working to improve stability and performance – on a program only a couple months out of beta. Arena is struggling to operate now, even though it is missing a number of critical features. Adding more cards – especially cards that involve more zones and interactions – will further tax the program and the computers that run it. Some potential new additions – like a friends list – may not have too great an effect on performance. Others, like real tournament support and drafting against humans, will be more impactful. For a decade or more, I have used the analogy of parts of the program being the cargo that needs to be hauled.  MTGO is a semi-truck: not flashy or fast, but capable of hauling everything you need. Arena, it seems to me, is a sports car. It is fast, flashy and sexy, but it isn’t going to be able to handle a serious load.

But no analogy is perfect. Maybe Arena can handle the load once Wizards loads it up with everything a real online Magic program needs. In the meantime, though, don’t uninstall MTGO just yet. 

“one million words” on MTGO. “4MWords” on Arena.


Welcome back! by Paul Leicht at Sat, 12/14/2019 - 05:28
Paul Leicht's picture

<3 Pete at having you back in the writing saddle however briefly that may be. Thanks for writing this.

In regards to the Matchmaking:D+ commentary: I really get what you are saying but I would mark this down as a major problem with Arena over all as a F2P version of the game. Fremium almost always disadvantages the new player. And matchmaking is really hard to do based on card availability. I have a fairly large arena collection and yet it is not on par with competitive accounts because I have not bought into the system and I have only a few gems. (100 I think, which isn't enough to buy a coffee much less a game.)

I have watched friends play who *have* large competitive collections and the vast majority do so because they have paid into the system. IE: It only really works for a mtg player if that player is willing to pay to play. Notice I don't claim that it is pay to win. Though I do think pay to win as a concept includes this pay wall problem. I stopped playing because I found similar issues with playing and also because I did not want to be stuck on the hamster wheel of "must play x games each day or I fall behind" problem which is *how* the free to play system keeps things in line with those who pay to play.

I will say I think MTGA is the most impressive video game version of the game to date. Still not on par with MTGO but that's a matter of time I am sure. It is my actual hope to never be able to say otherwise because I like enjoying my MTGO collection and I would hate to see it disappear in a digital vapor trail because MTGA took over the business entirely. I doubt there could be enough compensation from WOTC if that did happen. It would probably be the last I ever played of the game again. But as I said, solid and impressive compared to all past attempts.

5 fireballs like in the olden days!

Ranked matchmaking doesn't by Cheater Hater at Sun, 12/15/2019 - 14:31
Cheater Hater's picture

Ranked matchmaking doesn't take deck strength into account (and shouldn't). Unranked does try to do deck strength, but it's based on number of wildcards used.

you could be right by one million words at Mon, 12/16/2019 - 19:04
one million words's picture

But even in unranked, my deck full of drafted cards, decks for which I did not burn wildcards, was repeatedly matched against decks full of rares and mythics...