70% will win 70% of the time without exception. Just be glad you're actually attacking at high odds.
Now what I don't like is the complication of unit damage with the RNG - specifically attacking archers (or even regular units!) with something that is unlikely to win like a chariot (I have actually won attacking an archer fortified on a hill with a chariot, but I wouldn't be my life on it happening very often). The problem is, sometimes (RARELY) the chariot wins. Sometimes, it does damage and dies. Sometimes, IT DOES ZERO DAMAGE AT ALL. The presence of first strikes of course play a huge role here, but few things piss me off like seeing a combat I rifle emerge from 3 battles in a row unscathed. Worse yet is when longbows take no damage from horse archers or knights, which are immune to the first strikes and possibly the best units for doing damage to the super strong top defender in theory.
When first strikes become involved it starts getting very, very hard to gauge whether or not the RNG is "reasonably certain" to deal you a win. I'll say this though, I've come to respect both first strikes and the drill promotion. A lot. Enough that I mix in drill rifles with the combat ones, and the drill ones actually attack city units damaged by siege collateral (the combat ones are for knights/cuirassers and the really strong stuff).
Random seed on reload isn't the problem, it's reloading itself that is kind of rigged IMO. The incentive and ability to cheat are too high (as if reloading isn't a tad rigged as is - a lot of outcomes can be varied pretty drastically by re-arranging attack order or just movement actions).
For example, if you lost that great general at 97% odds, or a CR III unit, you could, knowing that would occur, reload even without random seed and instead send a crap unit for the "rng screws you" fight. I don't know the EXACT mechanics of the RNG, but let's give a simple example of no random seed RNG outcomes from 1 to 100:
25, 89, 1, 15, 73.
If your CR III unit got anything higher than a 5 it would win, but you attacked with it 3rd. Knowing that, you attack with a draft unit or some scrubby leftover chariot or something 3rd, and then CR III unit goes 4th, and since the 73 is pretty good you use a stock CR II guy next. Granted, you can't see these numbers outright, but you can guess, often pretty well...
The way I see it, the random seed just adds more variety, but reloading is "cheating" regardless, although that's kind of a weighted term. Sometimes, the game lags and your super medic moves 2 spaces into enemy territory next to a spear, the apostolic palace glitches because you killed the AP resident, etc etc.
Or, my personal favorite, your hand isn't even on the keyboard, but the computer thinks you're holding alt, and therefore when you click on your religious buddy's name for a trade at friendly, it instead sounds the horn. Anybody who claims reloading on THAT is cheating needs to relax and perhaps learn to apply logic. However, a lot of things ARE pretty grey area when it comes to this - I don't personally see random seed as making too big a difference, but I don't see how it hurts or "encourages cheating" - you can sure as hell cheat via reloads without it, and pretty much as easily.