I think many questions would be vastly improved by a concise single complete grammatical sentence that summarized the problem in the form of a question.

Some questions however, are still very tantalizing even without a question mark. Take this recent question by Jon Skeet:

• Curious null-coalescing operator custom implicit conversion behaviour

The title is not a question but a introduction to the question that will be asked in the body. Should the question instead be:

• Is the behavior of this null-coalescing operator in the presence of a custom implicit conversion curious?

I guess I like it the way he wrote it.

A bad title would be:

Suspicious operator behavior

and changing it into:

Is this operator behavior suspicious?

doesn't really improve it.

But of course, the worst is the mish-mash we usually see:

C# - Operator not working (compiler problem, conversion). Is this right?

The point being that the badness of the titles isn't usually related to their structure as sentences so much as the inability of the question author to compose an effective informative title that draws you in while staying within the bounds of convention.

At the moment that convention (in the field) does not require that the title end with a question mark nor be a complete sentence. While changing that convention would really improve a lot of basic questions, it would also limit the expressiveness of the title for others.

I think many questions would be vastly improved by a concise single complete grammatical sentence that summarized the problem in the form of a question.

Some questions however, are still very tantalizing even without a question mark. Take this recent question by Jon Skeet:

• Curious null-coalescing operator custom implicit conversion behaviour

The title is not a question but a introduction to the question that will be asked in the body. Should the question instead be:

• Is the behavior of this null-coalescing operator in the presence of a custom implicit conversion curious?

I guess I like it the way he wrote it.

A bad title would be:

Suspicious operator behavior

and changing it into:

Is this operator behavior suspicious?

doesn't really improve it.

But of course, the worst is the mish-mash we usually see:

C# - Operator not working (compiler problem, conversion). Is this right?

The point being that the badness of the titles isn't usually related to their structure as sentences so much as the inability of the question author to compose an effective informative title that draws you in while staying within the bounds of convention.

At the moment that convention (in the field) does not require that the title end with a question mark nor be a complete sentence. While changing that convention would really improve a lot of basic questions, it would also limit the expressiveness of the title for others.