Every mechanism has a dead point
Where a linkage loses the ability to transmit force.
Every mechanism built from pivoting links has at least one position where pushing on it, however hard, produces no turning effect at all, because the direction the force is being applied in lines up exactly with the pivot it is meant to be turning, leaving nothing left over to actually rotate anything.
The short version
A force only turns something around a pivot to the extent that it pushes sideways relative to the arm connecting them, and a force applied straight along that arm, directly toward or away from the pivot, produces no turning effect whatsoever regardless of how large that force is. A dead point is exactly the position in a mechanism's travel where this alignment happens, the connecting link and the direction of the applied force lining up so precisely that the lever arm available to produce a turning effect briefly shrinks to nothing, and a mechanism sitting exactly at that position simply cannot be started moving by the same kind of push that moves it easily everywhere else.
This is a genuinely different problem from a mechanism simply being hard to move, since a dead point is not about how much force is available, it is about geometry temporarily removing any way for that force, however large, to actually be converted into rotation. A far stronger push at exactly the dead point still produces exactly no turning effect, which is the detail that makes a dead point worth naming separately rather than simply treating it as an unusually stiff moment in an otherwise smooth motion.
The bicycle-pedal comparison
A cyclist pushing straight down on a pedal that happens to be at the very top or the very bottom of its circular path is pushing directly along the line between the pedal and the crank's own pivot, and that push, however hard, does almost nothing to turn the crank, since a force aligned with the crank arm has no sideways component left to actually rotate it. Anyone who has ever stalled a bicycle trying to pull away from a stop with a pedal sitting at exactly that position has felt a dead point directly, through their own legs, and the usual fix, easing off, letting the other foot's momentum carry the crank a little further round, or simply rocking the bike slightly to nudge past the stuck position, is the same fix every mechanical dead point ultimately needs, some source of motion from somewhere other than the force that is currently stuck.
Why momentum is what actually carries a mechanism through
A mechanism moving continuously rarely notices its own dead points, since the rotating mass already in motion carries enough momentum to coast straight through the position where the driving force temporarily has nothing to push against, exactly the way a cyclist's own moving legs and the bicycle's own rolling momentum carry the pedal past its dead point without the rider ever consciously managing it. This is precisely why a multi-cylinder engine stages its cylinders to fire at different points in the crankshaft's rotation rather than all at once, and why so many single-cylinder engines and hand-cranked machines carry a heavy flywheel on the same shaft, so that whenever the driving force happens to sit at its own dead point, the flywheel's own stored momentum, or another cylinder elsewhere in its cycle still actively pushing, carries the whole assembly straight through a position no single driving force alone could ever start moving from a dead stop.
The number that matters here
A mechanism started from complete rest with its crank sitting within a few degrees of true dead centre needs a noticeably larger initial push, sometimes several times the force needed to get it moving from any other position, purely because the lever arm available to convert that push into rotation has shrunk to nearly nothing right at that specific point. Moved even a small angle away from dead centre, the same mechanism regains most of its normal leverage almost immediately, which is why a stalled dead point is usually a brief, easily solved nuisance rather than a lasting problem, provided something can nudge the mechanism even slightly off the exact stuck position. This sharp recovery is itself the reason a small nudge, a light tap, a slight rock, is so often all a stuck mechanism actually needs, rather than a proportionally larger sustained force applied in the same stuck direction.
What this changes in practice
Designing a mechanism that has to start reliably from rest, rather than one that is merely kept running once already moving, means either avoiding a dead point falling exactly where the mechanism is expected to start, offsetting the geometry slightly so the stuck position and the resting position never coincide, or deliberately supplying an alternative source of turning force, a flywheel's stored momentum, a second staggered driving element, specifically to cover the moment the primary force has nothing left to push with. Ignoring the dead point entirely and hoping a mechanism will simply start wherever it happens to be resting is a reasonable gamble for a low-consequence device and a genuine design failure for anything that needs to start reliably every single time, since a mechanism that only fails to start once in every few dozen attempts is still failing often enough to matter for anything used daily.
What this does not explain
A dead point being crossed easily by momentum says nothing about how efficiently the mechanism runs on either side of that position, since a mechanism can pass smoothly through its dead point and still be a poor, wasteful design everywhere else in its cycle for entirely unrelated reasons. Solving the dead point problem specifically answers whether a mechanism can start and keep running at all, not whether it is doing so well once it has, and the next article in this set turns that same weakness around, showing how a mechanism can be deliberately built to stall on purpose, exactly at the point where a load needs holding rather than moving.