Atlas Predicts: Which party will win the 2016 presidential election?

[Mallow]

Also, the polling right now that says "33% of TRUMP supporters won't vote Cruz in the general!!!! (or vice versa)" is really unreliable because there will be incredible pressure to unify around the nominee once there is a nominee and the alternative is Hillary. Perhaps the L nominee gets up to 3% or so, but other than that the #NEVERTRUMP/#NEVERCRUZ movement (a lot of the new esablishment supporters Cruz has are just supporting him so TRUMP doesn't get to 1237 on the first ballot and they can try to nominate Ryan/Kasich/Romney instead) will largely fizzle out due to fear of Hillary and such people will just hold their nose for the R nominee. Just as in 2012, the race will be mostly decided on turnout and on the 7%ish of the electorate that consists of actual swing voters. You will NOT see Hillary getting 20% of the Republican vote.

That's all fine, but it seems like you're conflating "33% of Trump supporters won't vote Cruz in the GE" with "33% of Trump supporters WILL vote Clinton in the GE". You said it yourself--it's about turnout. And the #NeverTrump movement may be overstated, but can you really suggest it won't substantially affect turnout? I imagine it would have to.

[Mallow]

...

Finally, is the Democratic Wall really more EVs?  What's really on the table if we're going to assume it's an election with a popular vote within 3%.  What kind of Republican is running?  In this map, the Republicans' wall has more EVs than the Democrats':

R: 191
D: 186

You might scoff at this map, but I think that if we're talking about a close popular vote, this is a realistic battleground.

Michigan is definitely arguable. If it is a battleground, so is Oregon. Here is a list of the states that, in the last four elections, have had margins within 5 points of the national popular vote:
Nevada (D0.97)
Iowa (D1.46)
New Hampshire (D1.53)
Ohio (R1.80)
Colorado (R1.97)
Florida (R2.63)
Wisconsin (D3.07)
Pennsylvania (D3.30)
Minnesota (D3.67)
Virginia (R3.80)
New Mexico (D3.85)

If you want to also include states that have margins between 5 and 10 points, you'd add:
Oregon (D5.97)
Michigan (D6.33)
Missouri (R7.30)
Washington (D8.86)
North Carolina (R9.04)
Maine (D9.39)

[Mallow]

Michigan is definitely arguable. If it is a battleground, so is Oregon. Here is a list of the states that, in the last four elections, have had margins within 5 points of the national popular vote:
Nevada (D0.97)
Iowa (D1.46)
New Hampshire (D1.53)
Ohio (R1.80)
Colorado (R1.97)
Florida (R2.63)
Wisconsin (D3.07)
Pennsylvania (D3.30)
Minnesota (D3.67)
Virginia (R3.80)
New Mexico (D3.85)

If you want to also include states that have margins between 5 and 10 points, you'd add:
Oregon (D5.97)
Michigan (D6.33)
Missouri (R7.30)
Washington (D8.86)
North Carolina (R9.04)
Maine (D9.39)

Michigan is tricky, because I don't know what the recent population outflow does to the long-term voting pattern.  Missouri seems to be a lost cause over the long term, while North Carolina's demographics are shifting in the Democrats' favor.  Oregon's hipster Millennial population seems to have landed it solid D.

Trying to predict the battleground of a close popular vote is really hard, because I don't know what kind of Republican is going to manage to get the popular vote close.  A Republican unicorn.  Nationwide demographics and culture are now too much to overcome.

I don't entirely disagree, but my point is that very small differences in your definition of what makes a "battleground" can lead to very different electoral vote "advantages".

[Mallow]

I don't believe that's an even PV map, or close to one.  That's a Dem +3% map at least, in my opinion.  Throw the tossups to the Democrat, and you have Obama's 2012.  And PA isn't a tossup in this, despite its steady GOP trend since 2004.  In an even PV, PA probably is.

I think too much is made of PA's "trend". Compared to the nation as a whole, as you say, it has trended R since 2004 (that's only three elections), and only by 3.45 points of margin. States that were within 10 points of the national margin in 2012 have "trended" (2012 differential - 2004 differential) in the following ways:
PA: R+3.45
MN: R+2.12
NH: R+2.12
OH: R+1.25
FL: R+0.45
MI: R+0.29
IA: D+0.15
WI: D+0.23
OR: D+1.60
NV: D+2.94
CO: D+3.70
NC: D+4.07
NM: D+4.61
VA: D+5.75

So yes, by this metric, PA is the most strongly R-trending state of the 2012 "battlegrounds". However, I would argue that a better metric is a longer-term linear fit through the data, which smooths out some of the oddities from cycle-to-cycle. Here is the slope of the same list of states (change in margin per cycle) from 1992 through 2012...
MN: R+0.60
PA: R+0.04
IA: D+0.31
FL: D+0.44
OH: D+0.49
NC: D+0.70
MI: D+0.96
NH: D+1.10
WI: D+1.12
NM: D+1.31
OR: D+1.56
CO: D+1.58
NV: D+2.02
VA: D+2.32

I suppose one could argue that PA has almost the strongest R trend of the group, but the true trend since 1992 there is almost 0. It just so happens that most of the "battleground" states have trended D compared to the nation as a whole since 1992 (the strong R trends come almost exclusively from states which were heavily R in 2012 anyway). Of course, choosing a different starting year will change these results, so you could make the case for a stronger PA R trend starting from 1984 or 2004, but you can also get PA to have a D trend if you start in 1996. It's basically just a near zero longer-term trend there.

[Mallow]

I suppose one could argue that PA has almost the strongest R trend of the group, but the true trend since 1992 there is almost 0. It just so happens that most of the "battleground" states have trended D compared to the nation as a whole since 1992 (the strong R trends come almost exclusively from states which were heavily R in 2012 anyway). Of course, choosing a different starting year will change these results, so you could make the case for a stronger PA R trend starting from 1984 or 2004, but you can also get PA to have a D trend if you start in 1996. It's basically just a near zero longer-term trend there.

If PA, MI, and OH do not trend R, the Republicans do indeed have an electoral problem, because while electoral votes are being reallocated into R states (TX, UT, GA, AZ), Democrats are also flipping R/swing states - states that are also gaining electoral votes (CO, VA, NC, FL).

For a yin, there must be a yang.  But if Republican votes are shifting to suburban areas in heavily Democratic states, or growth in heavily Republican states, and are NOT to be found in the swing states, then they will still have an electoral problem.

The orthodoxy that "the electoral vote will all work out" is a myth.  The more we see states get polarized, the more we're going to see disagreements between the two in one direction or the other.

Agreed on all counts. Of course, one must be careful about extrapolating trends forward in time too far, as they can quickly change (see 1960-1992, which saw a number of states swing wildly back and forth several times).