Some equal temperament intervals are narrower than their just-intonation (nonbeating) counterparts, for example an ET perfect fifth (2.996614:2) and just perfect fifth (3:2). But others are wider, for example an ET perfect fourth (4.00452:3) and just perfect fourth (4:3), and an ET major third (5.039684:4) and just major third (5:4).
If you tune each string sequentially (low E to high E, or vice versa) and eliminate all of the beating each step of the way, the effect adds up to the point where it's quite noticeable, meaning that your low E and high E will sound like garbage when played together because that ratio should be precisely 4:1 but now you've accidentally made it narrower than that. How much narrower?
For a guitar going from low E string to high E string, we need to stack 3 perfect fourths, a major third, and another perfect fourth -- and end up at 4x the starting frequency. If we use those non-integer ratios of the 12 TET system (intentional beating), we end up with 4.00452/3 * 4.00452/3 * 4.00452/3 * 5.039684/4 * 4.00452/3 = 4.00000. That's what we want. But if we use the integer ratios of just-intonation (no beating), we end up with 4/3 * 4/3 * 4/3 * 5/4 * 4/3 = 3.95062 and that's going to sound like complete ass. This is why just-intonation is not used for instruments like piano and guitar that are designed to play in all keys. It's used by choirs and barbarshop quartets without piano accompaniment, since they can adapt on the fly, and it's glorious.
An electronic tuner is the most practical way to avoid this problem. Alternatively, you could just get the perfect fourths nonbeating, get the double-octave nonbeating, and let the major third beat however it wants -- this works because rounding the ~4.005:3 perfect fourth to 4/3 is somewhat acceptable but rounding the ~5.04/4 major third to 5/3 is not.
I had been thinking about listening for beating in the 5:4 and 4:3 intervals using open strings, which does present the just temperament problem, and harmonics on open strings to make a unison unfortunately shares this problem 100%.