An investigative role-playing game requires commitment from both the players and the GM to be a success. It is reasonable to expect more from a player than "I use my Spot Hidden skill" or even "I search the room." Players have skills just as their characters do, and role-playing is a far richer experience when they use those skills. Investigation should be described by the player in sufficient detail that anyone can imagine what the character is doing and the GM can ascertain whether any clues, if extant, would be discovered. It is also reasonable to expect more from a GM than "Roll your Spot Hidden skill." If a player has done what is necessary to be reasonably assured that a clue, if extant, would be discovered, then no roll should be required. If the clue is there and the effort is made to find it, it should be considered found. It's shocking in its simplicity.
The designers of Trail of Cthulhu may consider the rules they crafted a solution to the alleged problem inherent in Call of Cthulhu, i.e. "one bad die roll can derail an adventure," but that rests on the presumption that the existence of an investigative skill roll requires making an investigative skill roll. I would argue that these rolls, like other detection rolls in other role-playing games (q.v.), serve mainly as a second chance to make a discovery when a player mistakenly neglects to properly investigate an area through which he or she passes. As long as the GM and the players are doing their share of the role-playing and providing the necessary descriptions, then nothing can derail an adventure except the actions of the player characters.
As for the "solution" proposed in Trail of Cthulhu, I think never having to roll an investigative skill is just as misguided as always having to roll an investigative skill. In cases where player skill comes into play, such as describing how one searches a room, results should depend on neither dice rolls nor automatic success, but on the logical consequences of the actions described. If the player, having described the extent of the search, fails to describe an action that would lead to the discovery of a particular clue, but the character possesses a skill that would have a bearing on the search (such as Spot Hidden or another skill that applies), then the GM may make the roll secretly. If the pertinent skill is high enough or the clue obvious enough, then the GM may rule that the clue is found without making a skill roll. It all depends on the circumstances. Recognizing which rules to use, alter, or ignore in a given situation is really the basis of GMing itself.
If, on the other hand, a character possesses a specialized skill that the player does not have, then it is acceptable to assume that under ordinary circumstances the skill will be automatically successful, or that under extraordinary circumstances a skill roll will be required. In neither case will a detailed description of the action on the part of the player be necessary. In other words, one needn't be a doctor in real life to play one on T.V. (or in a role-playing game).
Character skills should extend the capabilities of the player, not limit them. Similarly, skill rolls should be made to extend the plot, not kill it. Even if a crucial roll results in failure, it should merely delay or misdirect the investigators [Edit: or allow them to succeed, but at a great cost]; it should not prevent them from reaching their ultimate goal. Only their ability as players should have a bearing on that.
[Originally posted in Fudgery.net/fudgerylog on 1 March 2012.]