Given the constellation of opinion on what should happen in January, and the arguments made by specific individuals in support or opposition to one view or another, perhaps it would be useful, instead of looking forward, to try to look backward.
We spend so much time and energy debating what should happen next, that we don't spend nearly enough time looking back to evaluate the arguments made in the past. This is an important task. If a person or group of people make arguments regarding a restriction that, with the benefit of hindsight, appear weaker than when initially presented or mistaken about some fundamental relationship, then it is necessary, as a feedback loop if nothing else, to call that out.
After last year's presidential election, there was much handwringing and retrospective analysis among the nation's pundit class about polling, modeling, and prognostications. No less should individuals who espoused, developed or articulated specific arguments about restrictions reconsider their positions at the time. Hindsight is 20/20.
Without purporting to comprehensively or systematically evaluate the spectrum of opinions developed over the course of the last few years, I will examine a few specific cases.
The pattern of restrictions over the last few years has provided a number of "natural" experiments with which to evaluate the various competing claims that arose in that time. That is to say, we can look at the proponents or opponents of specific restrictions, such as Golem, Chalice, Probe, Gush, Thorn and Mentor, with a good deal of evidence about what effect those specific restrictions had - or failed to have - in relation to what those advocates claimed. This includes the DCI.
The case of Gitaxian Probe is probably the easiest case. Here is what the DCI said about Probe:
In Vintage, the metagame has come to a bit of a standstill as Monastery Mentor decks face down their main predator, Workshop decks. The primary issue seems to revolve around the prevalence of free draw spells for the Mentor deck that let it churn through its library for no mana while creating an abundance of tokens. We believe by removing these free draw spells—and the perfect information that comes with Gitaxian Probe—we will significantly weaken Monastery Mentor–based strategies. Hopefully the move away from "free" spells in the Mentor decks will lessen the impact of the Workshop deck's various Sphere of Resistance effects, opening up the metagame.
We now know, with the benefit of hindsight, that both predictions were wrong. Restricting Probe neither weakened Monastery Mentor decks nor "lessened the impact of Workshop deck's various sphere effects, opening up the metagame."
In a contemporaneous podcast recorded that day, I specifically claimed that both predictions would prove to be wrong. Specifically, I claimed that it was "more likely that the restriction of Gush would either keep Mentor at the same level or increase it than it would decrease the prevalance of Mentor." Empirically, I have been proven correct. I will not recite the data here, since it's been so well documented elsewhere.
More revealingly, you can see the full spectrum of opinions - and specific people - who supported that restriction (and the restriction of Gush): http://www.themanadrain.com/topic/1147/april-24th-2017-banned-and-restricted-update-gush-and-probe-top-in-legacy
To call out a few specific comments, @boxian correctly called out the problem with their logic. In contrast, @KingLeovold notably expressed his happiness that Mentor was not restricted.
There are many other empirically testable comments in that thread.
One was whether "restricting Gush opened up "other blue draw engines."
We can look back at the data and test this claims.
The % of Big Blue & Blue Control decks in the January P9 Challenge were 5.4% and 10.7% respectively, 19% and 3.2% respectively in Feb., and 3.2 and 6.5% in March. That's an average of 9.2% and 6.8% respectively.
In the compiled May metagame report, however, the % of Big Blue and Blue Control was, by averaging all of the Challenges, was 7.5% and 3.8%. In other words, Big Blue and Blue Control, was actually less than the average of the three preceding months. Far from opening up the metagame, the restriction of Gush appeared to result in fewer, not more, blue decks in the metagame. If looking at the "May-July" report, the metagame representation average for Big Blue and Blue Control was 5.7% each. Again, below the Jan-March average.
Quite the contrary, these restrictions consolidated the metagame.
In 2016, there were a number of debates in various threads about whether Mentor, Gush or both cards should be restricted. I argued that Mentor should be restricted before Gush, because it was the card that was driving the metagame more than Gush. Matt Murray, felt that both Gush and Mentor should be restricted because they impacted the metagame in different ways. Regarding Gush, here is quote from last year that is representative of the argument that Matt repeated many times in 2016:
@chubbyrain said in Discussing Gush Mentor (beating it, restriction discussion, anything):
@ChubbyRain how good is thing in the ice? is it a metagame card or is it an actual threat that could stay?
My opinion has been that the actual win conditions in Gush decks are interchangeable and should be based on what you feel the metagame will be like. Mentor is arguably the most powerful in a vacuum, but people start running Sudden Shocks, Sulfur Elementals, Dread of Nights, etc. Thing in the Ice is a very powerful alternative. Shops becomes very popular...look to Delver/Young Pyromancer. Shops becomes nonexistent...look to Doomsday. People start cutting Dack Faydens and loading up on Supreme Verdicts...Tinker becomes viable. The format is actually very dynamic and open...so long as you are willing to run 3-4 Gush in your Blue decks. That's the one card I feel is metagame proof.
Although we lack data on the counter-factual of a contemporary metagame with Mentor Restricted, and Gush unrestricted, the argument quoted here has been, I believe, refuted or at least strongly undermined. That is to say, the argument that 'win conditions in Gush decks are interchangeable' is strongly undermined by the evidence following the restriction of Mentor. The evidence is overwhelming that restricting Mentor led to statistically large observed decline in Turbo Xerox (formerly Gush) strategies, but Gush did not.
Of course, it is possible that it took the combined restrictions of Gush and Mentor to produce this result, but I spoke with many people after Vintage Champs who agreed with me that unrestricted Gush would not have made a difference in the Top 8 result. The fact that TX strategies saw no decline after the restriction of Gush, but saw a large decline after the restriction of Mentor, strongly, but not irrefutably, supports the view that Matt's argument above was incorrect.
That is not to say that Gush 1) should not have been restricted or 2) would not have eventually been restricted. But, more narrowly, it is to say that the argument for why Gush had to be restricted presented here appears to be empirically false, based upon available evidence.
As the restriction of Mentor shows, the win condition really does matter. Young Pyromancer, Thing in the Ice, Hydra, etc. -- those cards are leagues behind Mentor in power level, especially in dealing with Workshops.
In summary, the performance of the "Gush" deck (or TX) after the restriction of Gush in April but before the restriction of Mentor in August, in comparison to it's performance since the restriction of Mentor is pretty strong evidence underming the argument developed in the quote above. Granted, Matt had other arguments regarding Gush, such as whether Gush suppresses other blue decks, but the performance of Mentor post-Gush restriction and the performance of tokens decks post-Mentor restriction really does underscore the power level of Mentor vis-a-vis the other win conditions. It certainly appears that Mentor was really anchoring those decks. And it's printing just a few months after Cruise/Dig created a huge boost to that archetype.
Regarding the argument that Gush suppresses other blue decks, given what we know now, I think the evidence is surprisingly (even to me) weak that it, and not simply Mentor, did that as well. After all, there does not appear to be a fundamental composition shift in the blue portion of the metagame through these restrictions that was not observable before them.
Similarly, re-consider Rich's theory about how Turbo Xerox decks "propped up" Shops. Specifically, in his conclusion in the OP, Rich argued against the restriction of anything from Shops, on this basis:
Workshop decks are looking like a dominant deck. However, as I've described above, I believe that this is because Workshops is one of the few viable ways to attack the Turbo Xerox Mentor deck. If anything is hit from the Workhsop deck right now, the only result would be to collapse and condense the metagame further. In other words, the strength of Turbo Xerox Mentor decks is causing Ravager Shops to occupy an outsized portion of the Vintage metagame.This is because Ravager Shops is the best response to Turbo Xerox Mentor.
Yet, empirically, this, too, has been proven false. The DCI ended up following his advice of restriction Mentor, but restricted Thorn contrary to his advice, and as the September aggregate MTGO data showed and the Vintage Championship results show, Shops were the strongest deck in the metagame terms of both prevalence and win %.
Although the theory that TX was propping up Shops was not unreasonable, it is another example, in my view, of a pre-restriction theory that has been empirically refuted by post-restriction evidence in several respects. Although the metagame did condense further, it was not in the manner that Rich warned, but rather towards more Workshops.
Regarding the general theory, every single restriction to the TX deck in the last few years has resulted in more Shops and better Shop performance, not less. After Gush's restriction, Shops surged to 40% of Top 8s on MTGO. Then Mentor was restricted, and we have Shops giving it's best ever performance at Vintage Champs.
Time permitted, I will dig further into the TMD archives to find other arguments that may be suspectible either to direct falsifiability based upon recent data or indirect refutation or undermining. This is an important exercise, as the guidance upon which B&R policy is made is sometimes based upon theories or models of the metagame that appear to be faulty or incorrect, especially with the benefit of hindsight. Consequentially, this may lead to suboptimal restrictions.
Without opining on specific cards that have yet to be restricted, I hold little doubt that we've suffered through a series of at least some mistakes where the DCI restricted the wrong card to address the target problem: Chalice instead of Golem, Gush & Probe instead of Mentor, and Thorn instead of Sphere. Even if we felt that Gush and/or Probe needed restriction, I think that everyone can agree that the DCI's logic proved faulty.
In each of these cases (or, at least 2 of the 3 so far), the mistake has required further action. It's likely, given what we know now, that restricting Golem would not have prevented Chalice from also getting restricted (since Thorn was axed, and more cards are likely coming). But that doesn't mean that Golem wouldn't have been the better initial restriction in late 2015.
Moreover, it's much less clear, and certainly plausible, that the restriction of Mentor could have forestalled or avoided the restriction of Gush and/or Probe. According the logic of the DCI in announcing the Gush & Probe restriction, that seems likely, since it focused entirely on Mentor, and the role that Probe and Gush played in amplifying Mentor.
Empirically, the restriction of Mentor has apparently brought TX decks in line. If you just line up the graphs, the restriction of Gush did nothing - as I predicted - to diminish Mentor's prevalence (or the popularity/success of TX decks), while the restriction of Mentor has led to a drastic decline of TX decks, from roughly 30% to roughly half (13.5% on MTGO).
It's possible that Gush would have still needed restriction, but the available evidence for that position is weaker than even I could have imagined.