Our contributor-dev has quite neat explanation.

A proof-of-burn system modifies the distribution of the state-transition probabilities in accordance with the number of coins which are “burned” by their owner. The process of burning a coin entails sending it to either a randomly-selected address which is unlikely to be under control of any participant or a fabricated address which is valid yet demonstrably uncontrolled. The amount of burned coins then serves as a ticket in a lottery for the right to publish state transitions. Participants can then engage in a proof-of-work-style puzzle whose difficulty is inversely related in proportion to the amount of burned coins. In some proof-of-burn systems the ticket is everlasting but in other systems such as Slimcoin, their effect on puzzle difficulty is gradually reduced to zero over a period of time, the burned coins degrade into “ash”.

1. Why generic transaction to a supposedly uncontrolled address instead of an introduction of a new special "burning" transaction type?

2. Complexity ladder has an exponential distribution (Every zero in the hash doubles the complexity, steps are discrete). Does that mean that your moneyburn will have exponential sweet spots in money spent?

3. How "difficulty" of the burn is adjusted and accommodated? Is it possible to hit "no zeroes required" threshold for the difficulty? What would happen when expected payout will reach that sum?

4. How would pooling work with such a combined mechanism? Would individual miners be able to contribute their moneyburn towards the pool's complexity limit?

Hi

How are you

1/ It's a special address for it, and special transaction - did you read whitepaper?

3/ Diff of burning is calculated on the number of burned coins, and in someway also alligned to POW diff.

No, I didn't, only your explanation. Unfortunately, I don't have a pdf-capable devices with me, since I'm on the trip atm.
That's why I've asked for the explanation

"No zeroes required" - I don't quite understand what you mean?

4/ No pool is required for such system - hash is so simple, that even rpi can resolve it in 90 sec block time

2/ I think that's not the case here, but I must contact with Graham to be sure.

I think first you should read whitepaper, because without reading it, we can't talk on the same subject.

4/ Pool is  only an option for POW related mining

As far as I understood, every contribution reduces the amount of zeroes for the pow part.
If "by default" the complexity requires, like, 10 zeroes, is it possible to reduce it ten times and just hash the transaction without any pow involved?

No it's not possible, every time burn block is added it must be surrounded by POW blocks back and forth. It's explained in whitepaper and code itself, I think.

I can make you contact with Graham, because he has better understand of system, and he is our contributor - dev.

Maybe Graham can just join our chat and answer the questions?
I think it would be better not only for me, but for all the observers in the chat as well, I guess..

He doesn't have telegram.

He is on bitcointalk, I can show you our thread and you can type your questions there.

Or slack