Because, while most audio op-amps such as the NJM2068 and NE5534 have excellent RFI handling in an audio setup, without audio-band detection below 20MHz or so, they start detecting with horrifying efficacy once you get past 50MHz. This is especially true in the GHz's where you have high levels of energy from WiFi/mobile phones, which are easily induced via standing waves up into the shielded centre conductor of most audio cables.
To stop your phono preamp from becoming an incoming call/text message pre-notification device, you need at least 100pF (for a through-hole design, a bit less for SMT caps with lower series inductance), of shunt capacitance (C1), behind a resistor (R1), of relatively low value to prevent excessive Johnson-noise contribution, to absorb VHF/UHF interference. If you leave this out of the in the interests of flexibility, the input is at the mercy of the environment and complaints will quite rightly abound.
You could, of course, increase the value of R1, and decrease C1 to a minimum of 47pF, but this just increases the effective input noise of the device.
Having a fixed input capacitance also encourages the user to shorten their phono leads, which helps to minimise setup issues. A good cable is a short one, I say!
