Yes it does. If you continue to use combo point generating attacks then you are hurting your dps because you are wasting those CPs.

Even with 0/5 FA its still worth it to FB 'when appropriate'. Those last 3 talent points aren't really significant for dps.

The very first conclusions of Feral by Night project was:

- RiP/rake must have 100% uptime => false
- FB have to be used with no more, than 35 energy => false
- FB musnt be used, if it leads to RiP/rake down => false

Im prety sure, that analyses of Simulcraft can only back up these facts.

And here these fails are repeated as a form of someones common sense. And these fails are also published in some of the most frequent websites of the most famed PvE guilds. Then loads of ferals spec for 0/5 FA and getting some ridiculous 4/5 fufor or "last stand" into. Im not trying to convice all for 5/5 FA, because somes are mangle bots. But telling to ferals, that not FBing is for their DPS increase is hard mistake. This is not discusion about geming ArPen or use softcap instead, where you can maybe present some common sence, because we cant currently simulate fight enviroment, like swiching targets or AoE. Cliping timers or shreding over 5 CBP are proven mistakes since Simulcraft or FbN including FB in their feral rotations.

In general, there is no way to make our rotation ideal. The question is, how to make best DPS with current RNG in current fight. If you have 5 CBP and "reasonable" RiP/SR duration left, just FB. Its DPS increase even with 0 ArP, 0/5 FA and 100 energy. The situation often described here "halp, I FBd and my RiP failed for XY seconds" is umpleasant.

But its not DPS loss, if RiP is not fallen for "too long". This is why dogma of 3 rules up here is untrue. FB is DPS boost even, if your RiP fall. You cant say here, that every FB leads to fallen RiP. There are 3 situation after FB, not one - the worst:

- RiP wasnt down after FB => rly big DPS increase
- RiP was down after FB for few seconds => still DPS increase
- RiP was down after FB for horror amount of seconds => DPS loss

All of these situations can happen no matter your skill (manytime I had during berserk chain of ten nocrit shreds etc). But in average, there are far more of situations, when FB is DPS boost. Thats why simulations favor FBing.

Surprisingly, this is what I thought to but running

vs one with FB cancelled out through simulationcraft,

I came up with the same relative DPS numbers using my own softcapped ArP profile which is quite a few pieces away from BiS.
9017 non-FB vs 9026 FB -- mangling
9186 non-FB vs 9194 FB -- manglebot

The difference may become more pronounced when we (or at least I) have higher crit % and hardcap ArP but as it stands, ~1.5 FB over 5 minutes is roughly equal to 2 more shreds.
Personally though, I've found that not FBing especially during a trinket proc is a large dps loss though many factors do change from fight to fight, week to week.

Don't be mistaken, there's still work to do on the SC profile with regards to CP usage. I think the SR rules are a little too simplistic right now, which is causing SR to be refreshed a little early. The default profile, while it's the current "max DPS" profile we've come up with, is still using CPs more conservatively than necessary to maintain its uptimes. There is probably room for more FBs.

However, it's important to put those discoveries by FbN in context. First, there's been skepticism by many in the community about that FbN from its inception, although it's been very popular from its addon. But many people have found that they don't always like its recommendations, and can improve upon its DPS numbers by ignoring it. This tells you, first-hand, that it's not executing an optimal cycle, but merely Nightcrowler's best formulation. Which was good, but not exactly perfect.

Additionally, all the research into FBing happened at the Naxx level-- which is forever ago, relative to current gear. At the time, I believe, the testing was done before the Primal Gore talent! We had more Rip uptime because of the 2t7 bonus, which changed how "safe" we had to be. I'm not sure which scales better, but Rip scales VERY well with gear.

That's not at all saying the concept isn't valid-- in theory, you should be able to use an FB when it will, on average, not cause more than X seconds of Rip downtime. The problem is, X is a lot smaller than it used to be, so FB is a lot less relevant. Which also means that the small gain in DPS from 5 talent points in FA is likely to be outweighed by survivability or utility talents. Which makes FB even less relevant.

The most important thing to remember is: if you're FBing, you have a lot more DPS to lose by FBing too much than by FBing too little. At most, you only lose 100 DPS by not FBing, whereas you can lose hundreds of DPS by letting your Rip uptime drop 20%.

I don't think that was Tun's exact criticism, Kaz-- I think he was discrediting all of the theorycraft for overestimating DPS, but just doing it in the FbN thread because it was the most active at the time. That, and that Tun has never used FB in his cycles, so he thought the concept had to be wrong in general. FbN was a simulation, though, and not a calculation like Rawr/Toskk, so it's unlikely he was FBing more than he possibly could; it's more that, for some reason, he was getting better answers with high FB usage than he probably should.

I wouldn't say any piece of theorycraft is getting things exactly right right now-- they all overestimate damage (in a better-than-an-optimal-raid way) and, to varying degrees, improperly estimate finisher proportions. Rawr and Toskk both overestimate FBs more than FbN did-- and Rawr has always been the king of overestimating DPS. They both assume 100% bleed uptimes with very minor exceptions; Rawr is somewhat fixable by artificially increasing the lag variable, but that's certainly not intended. Rawr also has no start-up lag. Even when playing to optimize bleed times, you rarely see a >90% on Rip, and almost never on Rake. Especially mangle bots.

The key to using any theorycraft is first understanding why it's wrong, and realize what magnitude its answer might be wrong.