I don't agree that reducing inflation benefits investors really. It is neutral.

There are basically three classes of usage rewards/inflation can be allocated to:

1. Positive (spending $1 results in >$1 contributed back to the value of the system)
2. Neutral (spending $1 results in exactly $1 contributed value)
3. Negative (spending $1 results in <$1 contributed value)

Burnpost is #2 (but see note at the bottom for an exception), it doesn't help or hurt anyone.

Now, since the pool is short term zero sum, when one upvotes in any of these categories, it reduces the others. But now you have to consider not only aggregate value but marginal value. This gets kind of complicated, so if there is enough interest, we can dive deeper into it elsewhere.

Ultimately, I think there should be an equilibrium where voters have to choose between #1 and #2; unless the system is very broken #3 should be driven to almost zero. The choice between #1 and #2 depends on marginal value, not total value. And FWIW, I do not think we are in a situation where #3 has been driven to almost zero, which is to say I consider the rewarding function to be very broken, still.

To that point, most of the recipients on this list, who post day after day and get rewarded day after day either by self-voting, vote trading, bought votes, or whatever other manner, are not the tail. That is to say, again, I do not believe whatsoever that #3 has been driven to near-zero. The system is in fact, still pretty broken (though somewhat less so than before EIP).

Some other points of (mild) disagreement.

1. Self-voting does not stay with investors (plaural), and is very, very harmful. It stays with the particular investor (singular) who does it, paid for almost entirely by the rest of investors. This is a key difference. When one investor self-votes, other investors are now in a prisoners' dilemma situation where they already see the other prisoner defecting. This ultimately forces all investors into choosing between self-voting (and vote-selling, vote-trading, etc. which are all equivalent) or suffering losses, and was essentially the path we were on prior to EIP. The system then collapses, with no rewards left for the tail. By contrast, voting for burnpost is perfectly compatible with the same or other voters also voting for a category 1 tail, without anyone needing to take losses to do it. Also, this perfectly fits with your model of "what if everyone did this?". Everyone can make some votes for burnpost and some votes for category 1 contributions (when seeing an opportunity for marginal gain there), and all would be fine.
2. Delegating to 'projects' is not a panacea whatsoever. Some of them might be good, but there have surely been, and continue to be, 'projects' which are ultimately disguised self-voting (i.e. very, very harmful), or just dumb ideas which don't add value.

Final note: In the situation when SBD is worth more than $1, burnpost shifts into category 1. That's because 1 STEEM gets converted into 1 STEEM worth of SBD (assuming SBD to be worth $1) via the payout code, which then is traded by burnpost into more than 1 STEEM, which is then burned (returned to the system/stakeholders). In this instance, there is a net gain. That doesn't mean there won't be other potential payouts which are category 1, value adding to a greater degree (and therefore more worthy of votes than burnpost), but the bar gets higher and higher as the value of SBD increases above $1.