@smmenen As you well know, the issue then becomes that any subsequent printing can undo the effect of the narrow restriction and push Shops back to a problematic place. You then have an unnecessary card on the restricted list (the one you hit to tailor) and the need to add another.
And this is arguably what happened when Walking Ballista (or Fleetwheel Cruiser or Thought-Knot Seer) was printed. The meaning of the Chalice and then Lodestone restrictions was destroyed by new printings. To restore the meaning of those restrictions, you had to narrowly tailor again, by adding Thorn to the list. We now have add three Modern-legal artifacts to the Vintage restricted list in as many years and are discussing adding another.
While I agree the strategy of narrow tailoring can be effective at achieving desired metagame percentages in the short term, it leads to restricted list inflation in the long term, as we have seen.
While a very plausible claim, the problem is that it isn't necessarily true. While there are many cases in which that claim has proven true (Necro vis-a-vis Dark Ritual in old Extended and Mentor vis-a-vis Gush last year), there are actually plenty of counter-examples or cases where this claim was made and it turned out not to be true. So, there is no neutral and objective a priori way to distinguish between times where it will be true and times when it won't. The short term view is actually all that matters, because the long term (the future) is fundamentally unknowable.
Also, there is problem of chronology. There were calls to restrict Workshop in 2003 and 2004, especially during the Trinsphere era. That actually presented a pure version of your problem: (that Y is the 'real problem', and if you restrict/ban X, you will ultimately have to restrict/ban Y), where Y = Workshop, and X=Trinisphere. But consider the fact that there were no other Workshop restrictions until 2015. That was an 11 year period where, arguably, no restriction was needed to deal with Workshop. Based upon the interval period, I think it's fair to say that restricting Workshop in 2004 instead of Trinisphere would have been the incorrect decision in relation to format diversity, which is my top line criteria.
I agree with you that the restriction of Golem has been swamped by events and recent printings. But a neutral and objective policymaker who was only looking at data could not know that at the time (even if the DCI were R&D members and tested a FFL Vintage, they still couldn't 'know' with confidence).
Assuming you (or anyone) believed that restricting Gush last year was necessary, does that make it 'wrong' to have unrestricted Gush in 2010? Maybe Nick Detwiler thinks so, but I think the fact that Gush was essentially under 15% of the metagame from 2010 to the summer of 2014 shows that it was the right decision in 2010.
Again, I don't think that restricting Workshop is unreasonable. I've been on record saying so. But I'm not sure that restricting Workshop would even solve the problem, since these Workshop decks lean less on Workshop than any other Workshop deck in the history of Workshop decks in Vintage since they are so low the ground, powering out maindeck spells that can almost all be cast for 2 mana.
But, most importantly, yes, you are correct in the general sense that, as I already said, a narrowly tailored policy approach has a higher risk of "failure" in the sense that more restrictions will be required than a more sweeping approach. That's a given. But there is an important trade off. While it may require more or deeper restrictions in the future, a narrowly tailored policy approach dramatically reduces the risk of unnecessary restrictions.
In my view, the risk of too many restrictions is a greater harm than the risk of insufficient restrictions, in terms of my goal of promoting overall strategic diversity. Once a card is taken away, it's much harder to get it back then taking away another card to deal with the original problem. To prove this, one need not look further than the fact that before 2015, every single instance in which the DCI had restricted more than 1 card, it later needed to unrestrict a card (with the only exception being Necro and Consult in 2000). For example, in 2001, they could have just restricted LED instead of also restricting Burning Wish and Chrome Mox.
While I don't think it is valid to speculate about what might happen in future when deciding what to do now, I do think there is one additional point that could justify choosing Y over X: If you discover that a long incidence of restriction debates involving Y, and also conclude that restricting Y instead of a new X would actually allow you to unrestrict other cards. So, if the DCI decided that restricting Workshop now would allow them to unrestrict, say, Chalice, Thorn and Golem, then I think that may tip the balance. So that would require two steps: both that card Y has contributed to several other restrictions, and that restricting Y would not only preclude restricting a new X, but that it could also allow an overall reduction in the number of cards on the restricted list. If only one of those conditions were satisfied, I would find that insufficient.