It works for UART where you have one to one Tx/Rx and it is enough to cross connect them ("data in" from one to "data out" from another") but in case of SPI where you can have multiple slave devices, all slaves "data in" are connected and are crossed with master. so that you need to differentiate which one is which.

Because "Data in" depends on perspective. With spi you connect mosi to mosi, on other naming schemes you connect TX to RX because for one side its sending data and on the other side its receiving it.