why does the solver want to shove 320% pot on the flop?

All options with the check mark are within 0.01 EV. Even the other 2 options are within 0.02 EV. So at GTO, none of the options are a huge mistake in EV.

The solver is very meticulously calibrated though. It is going to have the optimal value/bluff combos in the jam range and the response is going to be optimal, including mixing with the right frequencies with indifferent hands. One of the reasons solver will mix with the huge size and even prefer it here is in theory it allows us to bluff more, which maximizes our EV. We have so many hands that have a ton of equity, including unmade hands like ours that probably has over 60% equity vs big blind range, and up to 10 outs + backdoor spades when called.

In practice vs humans who cannot play GTO, one of these options is probably going to be best vs humans. We should generally try to pick the maximum exploitative strategy vs humans. Short of picking the perfect exploitative size for our hand vs a player's range, we probably want to simplify the solver strategy into something that is more implmentable by humans. IE, you pick one size for this board texture, either smaller bet with a higher frequency or larger bet with a lower frequency.

I would imagine we don't have to implement 3x pot flop shoves to have a strong strategy in practice.

I like the shove. In practice I have done this (and lost but it was Prague and I was the only one with 18 bb's and my opponent had 2 pair).

The issue here is that the board has JT in it. That means that there are a ton of straights that hit if we check or bet small (and are called). Yes we hit one of them but BB can hit the rest. So a 7, 8, 9, Q, A could be bad for us as would be a spade. Roughly over half the cards in the deck.

The other issue is if we bet 2+ bb's the effective stack size means that BB will shove all c/r's. Are we going to be calling? I don't like it really because I think BB would shove with some hands that would have folded to our shove. If they are balanced then we are in a -EV situation.

If we bet like 1 bb (the 16.4% solver bet) we have no FE vs any Jx, Tx, and 3x hands and any draws. This is a bet size I just don't like. It is basically an in position blocking bet which says we will likely be folding to a c/r especially a jam.

The shove gets a lot of draws to fold and the ones that don't fold we are likely ahead. It might get Tx and 3x hands to fold. It may look like an overbluff action but hands like PP's <= 99 will likely fold. And if BB has a Jx or better hand then so be it.

In the end the reason for the jam is that the more we are the aggressors the higher our chances for winning as long as we are balanced (and this is probably why its a 53% thing Solver-wise)

It’s funny, I’ve seen and heard about this spot before, and I was guessing we’d see a JTx flush draw board before I opened the thread.

Yeah, I thought the same thing before I opened the thread. JTx flops are to my knowledge the only flops that include massive shoves as a major part of your strategy.

I think it probably is the best play, even against humans. People are going to make huge errors against that strategy. That being said I haven't incorporated it into my game yet. You've got to be very careful to use the right combos and frequencies otherwise it's going to be easy to make massive blunders.

A few other advantages not mentioned by others yet:

When we have hands like over pairs shoving allows us to get max value against his top pair type hands before scare cards hit that allow him to get away.

You force him to fold out a lot of hands with major equity, which is a win for us.

It's been a while since I looked at the spot, but I remember being somewhat surprised how wide he has to call. You force him to call with a lot of hands that don't really want to stack off, otherwise you'll print with your semibluffs like the 89 pictured here.

Risking 18 to win 5.5

At some point, you just have incentive to take the obvious tradeoff/gamble that this bet embodies:

1. risk the max right now, in exchange for
2. max immediate win, which has reached a sufficiently large % of effective stacks to incentivize going for this
3. max protection for most vulnerable combos (really closely related to #2)
4. max equity realization for combos that benefit the most--draws to whatever, they just have to be sufficiently strong when called.

It's not the strongest not the weakest combos that benefit from mixing in this strategy:

Eventually, at like 25bb the incentive to jam goes away. It's too expensive a proposition for what you're purchasing:

• a guaranteed gamble for those combos which, even when called, win enough (have enough equity) to justify this
  
• max protection for most vulnerable combos that, even when called, win enough to justify this
  
• a lottery ticket where, hook or crook, you're gonna see all your numbers.
  

QQ, Jx, AK, AQ, KQ

The ones that need most protection and/or still do "well enough" when called and have no issues with buying itself guaranteed max equity realization.

You'd probably see a little some of it on QTx too

You don't see it on QdJs3s because Qx is much more valuable than Jx and needs less protection. The incentive structure is too lopsided on this texture to justify paying 18bb for that protection. In other words, you need less protection. But EVs might be close--this case is right on the edge.

You don't see it on Ts8d3s because you don't buy enough protection. Tx are the most common of the logical value choices: they need the most protection. Tx prefers having just a little more flexibility here: sometimes checking turns, sometimes checking back rivers. EVs might be close--this case is right on the edge.

It's the force of protection that drives things here.

It's not like Axx where your top pair needs minimal protection.