To find an optimised solution in an NP-hard problem computational power is needed
And as I've already demonstrated, miners choose to go with the "good enough" solution, rather than the ideal. It is simply not worth the effort to find the ideal solution with the given cost.
Do you think your rapspi will be able to handle a massive amount of paths and liquidity if/when LN is implemented ?
Lightning
is implemented. The algorithms used are as I've already said optimized not for the cheapest, but for the cheap enough fee. It will work on even larger network, likely less accurately.
But still people in here claim that it will work
There is a variety of opinions about LN in this board, and I highly doubt we all believe it will work flawlessly. There are tradeoffs, just with everything.
LN is even more centralised than mining . Have a look at how much liquidity and channels belong to the top 10
Now, who's arbitrarily using the word "centralized"?
