We would still have ASICs by now, and mining might be more centralized.

ASIC-resistance is a mirage. A device specifically engineered to only do one thing will always beat general computation devices. ASIC-resistance just makes it harder and more expensive to design one. Yet, as the exchange rate of Bitcoin increased, making one would have at some point been economically viable for someone. That manufacturer would then have had a monopoly on dedicated mining hardware and easily cornered the entire hardware market or directly the mining market. It is actually in our interest that the hashing algorithm is ASIC-friendly. It makes it easier for multiple producers to operate in the market and dedicated devices being the only viable hardware for mining means that miners are literally invested in the success of the Bitcoin project. Their hardware is useless for anything else, they cannot hop to other currencies as the exchange rates fluctuate or bet on multiple horses. An ASIC-friendly network also does not need to contend with sudden hashrate spikes from botnets turning their attention at the network and leaving just as suddenly.

Either way, the downsides of ASIC-resistance clearly outweigh the upsides. Introduction would have taken a hard-fork, disenfranchised a set of stakeholders in the Bitcoin network overnight by obsoleting all existing hashing investments, and yielded only uncertain benefits.