Although both the PHB and the Basic Rules suggest that these details will vary according to the DM's preference, the Monster Manual states clearly that sharks frenzy against any creature that doesn't have full HP.
The distinction between HP and actual wounds has never been clear, in any edition, but the fact that HP damage necessarily draws blood in 5E is irrefutable.
It's designed to be easy to play, by making certain basic assumptions, such as that the shark is fighting one or more of the PC races. In most cases, the shark isn't fighting a golem or skeleton, and it won't be fighting someone who has only suffered psychic damage. That's fair. If the basic assumptions don't hold, for whatever reason, then the DM is there to adjudicate that.That's a good argument, but hardly "irrefutable".
Bear in mind that damage can be psychic. It's pretty easy to argue that 1 point of psychic damage won't draws blood. In addition, a shark gets advantage on creatures that don't even have blood, so long as they are below maximum hps. A damaged water elemental? Advantage. A damaged stone golem? Advantage. A damaged skeleton? Advantage.
Sorry, I don't fully buy your argument. Given that there is no hard and fast rule for when you actually leak blood after being damaged, and given the above, I have to conclude that the shark's blood frenzy ability is designed to make it easy to play, not to model when damage causes a creature to bleed.
That being said, the general underlying principle of the mechanic is that - barring exceptional circumstances - any amount of damage to a PC will necessarily draw blood.
Every hit is a scratch. Just a slight one. Enough to deliver poison or break concentration.
If you go that route, you are usually safe.
You can break concentration without drawing blood, and you only need a scratch IF you have a creature with poison. The need for damage for poison and similar attacks is why the rule says "typically shows no signs of damage." Otherwise, 5e very explicitly does no physical damage at all until you drop below 50% hit points.
Every hit might be a scratch.
A scratch is a sign of injury, so according to RAW most hits do not result in scratches. They use "typically," because sometimes you need injury at over 50% hit points. You can always house rule every hit to show signs of injury, but as 5e is written that does not occur.
Now that I have over 50 votes (53), let's see what is happening:
Vote option 1 (23 votes): 0-1 times equates to a +15 modifier over +0
Vote option 2 (10 votes): 2-3 times equates to a +10 modifier over +0
Vote option 3 (12 votes): 4-5 times equates to a +6 modifier over +0
Vote option 4 (5 votes): 6-7 times equates to a +3 modifier over +0
Vote option 5 (3 votes): 8-9 times equates to a +1 modifier over +0
The expected value (using the mean for each voting block) is 2.8 times (within option 2) so a +10 modifier (with rounding, this is discrete after all).
Now, originally I was only comparing "proficiency bonus" against no proficiency bonus. As many people have pointed out, my word choice might have mislead some of the voting choices with the poll question, but hopefully the explanation I gave in the OP went into enough detail to remedy this. With the added caveat about not considering expertise, which would otherwise double the proficiency bonus.
Either way, if you understood my intent in only considering proficiency bonus (all other things being 0 or equal between the two contestants), then the +6 currently used in 5E doesn't really offer enough of a difference between maximum proficiency bonus and no proficiency. A difference that more represents the expected results of such a contest would be +10.
I know bounded accuracy was meant to stop the treadmill effect, and it has nicely, but personally I always felt there was not enough of a difference between the two ends of the spectrum. I don't want a full 20-points like in prior editions, but 6 seems too little. We've been playing with a house-rule that maxes out proficiency bonus at +8 for a while now, and considered raising it again.
All that being said, if you think that ability modifier must be included due to the intent of the designers and voted with that in mind, then the current +11 max (+5 for ability, +6 for proficiency) is close to the poll results and works pretty well.
Thanks to all for voting and adding your comments. It has given me much to think on.
Yes, seems to work out as expected.
Expertise or 20 in a stat seems totally right as ultimately skilled.
Well, to quote a well-known individual: "From a certain point of view."
See, unfortunately the problem is I thought I had made my interest clear that I was only concerned with the proficiency bonus itself, and I know some people voted with that in mind. However, because I used "ultimate skill", it was pointed out that it should, in fact, include the ability score modifier (which I wasn't).
Since I have no idea who voted with which schema in mind, I can't draw really great conclusions from all this.
It works out like this RAW:
If two people both have the same ability scores, the person with highest proficiency (sans expertise) of +6 would lose to a person with no proficiency at all 22.75% of the time, or roughly 4-5 times out of 20 would be expected (of course it can vary from trial to trial). Which is saying a Level 1 character would be able to beat a Level 20 in a skill contest (assuming no expertise, which most characters DON'T have, and the same ability scores) 4-5 times out of 20. To my mind, that makes the potential of 20 levels and hundreds of thousands of XP difference not as great as it should be.
Which is why our table increase maximum prof bonus at Level 20 to +8, and we will likely increase it a bit more... maybe +10--I don't know yet and have to discuss everything further with the DM and our table.
If you deliver poison it needs to be a scratch. Might be tiny. Might even be invisible, but you need to connect.
[You can break concentration without drawing blood, and you only need a scratch IF you have a creature with poison. The need for damage for poison and similar attacks is why the rule says "typically shows no signs of damage." Otherwise, 5e very explicitly does no physical damage at all until you drop below 50% hit points.
Do you even read before you respond to a post? I said this in the post you first responded to.
RAW is that there is typically no damage of any sort, which includes scratches. Typically = majority, usual, etc. A poisonous creature using it's poison attack type will be one of those atypical attacks that would need a scratch when the PC is above 50% hit points. Short of one of those atypical attacks, there is no scratch at above 50% if you are playing by RAW. That's how hit points work in 5e.
You mean this part?
"DESCRIBING THE EFFECTS OF DAMAGE
Dungeon Masters describe hit point loss in different ways. When your current hit point total is half or more of your hit point maximum, you typically show no signs of injury. When you drop below half your hit point maximum, you show signs of wear, such as cuts and bruises. An attack that reduces you to 0 hit points strikes you directly, leaving a bleeding injury or other trauma, or it simply knocks you unconscious."
That is not a rule, but a sidebar. It starts with:... Dungeon Masters do XXX, so it just describes how it could be done, a suggestion.
Then it follows with "typically", which can mean otherwise. So it might be a scratch or might not. If it is delivered by poison it needs to be a scratch, but other hits can also be narrated as a scratch.
It then follows with "injury". Yeah, that really depends how you define injury. I would not call a little scratch injury, but your mileage might vary.