Results on the thermal transition of QCD with 3 degenerate flavors, in the lower-left corner
of the Columbia plot, are puzzling.
The transition is expected to be first-order for massless quarks, and to remain so for
a range of quark masses until it turns second-order at a critical quark mass.
But this critical quark mass and resulting ''pion'' mass disagree violently between Wilson
and staggered fermions at finite lattice spacing, and decrease sharply with the lattice spacing,
for staggered fermions at least.
To clarify this puzzle and eliminate potential systematic effects from rooting, we study
the 4-flavor theory with staggered fermions, on lattices with 4 to 10 time-slices.
Our results are qualitatively similar to the 3-flavor case, so that rooting is not
an issue. However, dramatic cutoff effects are visible, even on our finest lattices.
Universality implies that cutoff effects for Wilson fermions are even more dramatic.
In order to obtain a first-order thermal transition in the continuum theory,
extremely light quarks are needed.