Everyone treats the butt joint like a phase you pass through. You start there, you learn dovetails, you stop coming back. That's backwards. The butt joint isn't where you start before you know better. It's what you reach for when you've thought carefully enough to know you don't need anything else.
The structural reality first: a butt joint has zero inherent mechanical resistance. End grain to face grain, no interlocking geometry, no racking resistance. Glued end grain bonds at 30–50% of long-grain efficiency under ideal conditions and approaches zero under cyclic tension. Without fasteners or reinforcement, it will not stay. That part of the argument against butt joints is accurate and complete.
What the argument misses is everything that follows. The butt joint's strength comes entirely from its reinforcement, which means the reinforcement can be precisely specified for what the connection actually needs to do. A pocket-screwed face frame doesn't need mortise-and-tenon strength. A screwed-and-glued cabinet back doesn't need dovetail-class tension resistance. The joint is adequate when the reinforcement is adequate, and adequate is all you ever need.
Reinforcement matched to load
Pocket screws are the production answer for face frames and light casework. One or two screws draw the joint, clamp it during glue cure, and provide moderate pull-apart resistance. Fast, accurate with a jig, sufficient for what face frames are asked to do. Not sufficient for chair rails or table aprons — anything that racks.
Screws and glue together handle most shop furniture and utility casework. The screw clamps while the glue cures; the glue provides adhesion on the long-grain faces in contact. Combined, they're adequate for fixed casework under static load.
Dowels add long-grain material across the joint plane. Better than screws at resisting pull-apart under repeated loading — no threads to work loose. They require precise alignment, which means a jig. Worth it when the connection needs to stay tight over years of use.
Biscuits are alignment tools. A biscuit adds negligible structural resistance once the glue dries. If you're thinking of a biscuit for strength, stop thinking that.
A floating tenon is a different animal entirely. It's a loose mortise-and-tenon, and you should call it that. It provides M&T-class resistance and you analyze it with M&T logic. At that point you're not building a butt joint — you're cutting a mortise by a different method.
A spline at a mitered corner solves a specific problem: both faces of a miter are end grain, which is a poor glue surface. A long-grain spline glued into matching grooves provides what the miter lacks. This is the structural answer to the mitered corner.
Where it belongs
Face frames. Pocket screws, adequate, fast. The cabinet carcase carries the structural load; the face frame just needs to be flat and attached.
Cabinet backs. Plywood nailed or screwed into a rabbet. The back is a shear panel and the fasteners are sufficient. If the rabbet is doing structural work, the rabbet is the joint — be precise about what's carrying what.
Shop furniture. The router sled, the finishing stand, the workbench torsion box. You know the load, it's static, and pocket screws with glue are correct.
Painted and hidden work. Substrate joinery where the finish covers everything. The joint doesn't need to prove anything.
Where it doesn't
Chair frames and table aprons. Racking is the primary load and a pocket-screwed connection will work loose under normal use. This is the most common expensive mistake from extending butt-joint logic too far. Any connection with cyclic, dynamic load — kitchen cabinet doors, workbench tops under mallet use, outdoor furniture — needs geometry. The screw thread works loose under repeated stress in a way that M&T geometry does not.
One more thing about glue
Gluing the end-grain butt face adds minimal strength. The long-grain faces of adjacent stock, if they're in contact, can be glued effectively. The end grain you're butting against cannot. In a pocket-screwed face frame, the glue on the long-grain stile and rail edges is working. The glue squeezed into the butt face is not. This isn't a reason to skip glue. It's a reason to understand where it's contributing.
The rule: the joint doesn't need to be stronger than the load requires. A dovetail on a cabinet back is not superior craft. A pocket-screwed chair rail is insufficient craft. The governing question is what the connection is actually solving.
There is no joint that is always wrong. There is only the wrong joint for the load.