Apples to Apples you are combaring this to other non-Concentration 1st level spells. So affecting two different rolls is actually a mild Strength, not a weakness.
Sure, that is why I put lack of concentration as a strength. that is a huge benefit of SB.
The mathematical effect on the battle you get for using a first level slot is smaller than the examples I said though. It is smaller than other non-concentration spells like shield and typically smaller than absorb elements too. You are spending a spell slot to get a battlefield effect and in that respect I don't think it keeps up with the spells I mentioned. That is what I am comparing in the text you quoted.
Umm, no. If Silvery Barbs gave a disadvantage or an advantage, it would have versatility. But it always offers both, so there's no versatility, there's two effects. If you compare, you cannot just discount away that the other affect will still happen and claim that it isn't as effective. It's very effective, just at two different things. Shield is very ineffective spell at giving a bonus while Silvery barbs is less ineffective at causing a miss - net effectiveness is absolutely for Silvery Barbs.
It is versatile in the way you can use it. You can use silvery Barbs against an attack targeting you, you can use it against a save by an enemy, you can use it on a skill check. You can use it to try to give someone else advantage on their attack (although this can be difficult to pull off in play since you don't choose when to use the advantage). That is why it is versatile, it can be used for all these things and can do some in tandem but it is not as effective at those things as other options.
Like you said silvery Barbs is less effective at causing enemies to miss you. If that is your goal shield is a better spell to use, it is more "effective" for that. SB is quasi effective at giving yourself advantage on your next attack, but truestrike will do that every time (unless you lose concentration).
Not as reliable - what is your baseline? For example you were comparing against Shield and Hex. Hex is disadvantage, so exact same reliability. Shield RAW you don't know if the AC bonus will turn the hit into a miss, so reliability is still unknown. Actually, since the is the chance a 20 which a Shield can never block it's probably more effective. I don't see any practical difference in reliability in those cases.
Hex causes disadvantage on every roll made with that ability while it is up, silvery Barbs can only be used on a roll the enemy succeeds on, only if you see him when he succeeds and if he is in range when he succeeds. That is why it is more reliable for this specific use, it will also affect far more rolls.
Shield is flat better at causing enemies to miss you, if someone hits you and you cast shield, the chance of not taking damage is higher than if you cast silvery barbs.
We can evaluate the reliability of shield because while we don't know the roll, we do know it hit, meaning it is from a smaller population than 1-20. Given this known (the attack hit), even if there is only 1 enemy attacking you with a single attack, shield is still more effective. If the enemy needs a 12 to hit you
and you don't know what he rolled but you know he hit; if you cast shield will turn that hit into a miss 62% of the time. Silvery Barbs cast in the same circumstance will result in a 55% chance of a miss. If the number needed to hit is 14 shield will stop that hit 84% of the time, while Silvery Barbs will stop it 65% of the time. If the number needed to hit is an 8, shield will stop this 42%, SB will stop the hit 35% of the time.
That is for a single attack too, while shield will continue to carry forward on multiple attacks, making it far more effective for this.
The only time SB is better is when the enemy scored a crit, or if you know the roll ahead of time and know it will beat shield (which is not the correct way to play as you said) and even then it might not be better if there are multiple enemies attacking you.
But then, it's actually twice as reliable. Since it effects two rolls. The chance that one of the rolls will end up better than giving (dis)advantage to just one roll is twice as good. So you can put this into the Mild Strength, that it is more reliable in total effect than than similar spells.
I disagree with this, partly because the second roll is not easily targeted, it is the next d20 rolled.
To be clear when I say "reliable", I am not talking about rolling dice I am talking about changing the effects on the battlefield.
The advantage is nearly random in play. The situation has to be right for it to be tightly controlled and you may end up using advantage it when it is either uneeded or irrelevant, making it "unreliable" in affecting the game. For example, my bladesinger used silvery barbs on an enemy attack and gave the Rogue advantage hoping to have him use it on sneak attack, then the Dragon breathes on us and the Rogue needs a 8 to save and only takes quarter damage if he fails .... and now he has advantage on that roll instead of his attack roll because it was the next d20 he rolled. Or the DM has all players roll perception to see if they notice something - that advantage is now used up. Or the worst case, I give advantage to myself and then have to roll a concentration save and I already have advantage on them, but it is the next d20 rolled, so it is gone. Those are actual examples from play. Most likely the advantage has little affect on that fight, in that respect it is not "reliable". Don't get me wrong advantage is always good to have, but it is not always effective, especially when you don't decide when to use it.
The targeted roll is more reliable, but the enemy already succeeded once. IME with this spell (which is now extensive), more often than not the enemy is going to succeed again on the attack/save/check.
So when I say it is not reliable I am talking about the effect on the battlefield. The vast majority of times you have the opportunity to cast it the outcome for the enemy will be the same and you will have a random advantage on a roll you can't control.
You can ration its use to make it more reliable - only use it when you have strong reason to believe it will change the outcome or when the order makes it easy to predict or perhaps even control where the advantage will happen, but then it is conditional.
