Why do I abandon things right before they work

It's almost there. One last step, and you suddenly let go — you stop tending to it and drift toward something new. By the time you look back, the thing that was about to take shape has gone cold. You're not lacking perseverance; the second it's "almost done" is far scarier than "just starting."

The unfinished thing is the safest thing

As long as it isn't finished, it lives in your imagination: maybe it works, maybe it's impressive. That possibility is soft — you can interpret it any way you like. But the moment it's actually done, it drops from imagination into reality and becomes a thing with a shape, something you're responsible for. Many people unconsciously protect that "untested potential," so they set it down by hand right before the test. Not because you don't want the result, but because you can't bear to lose the safe zone where anything is still possible.

Once it works, you have to carry it

A more practical fear: if it really works, the hard part is after. Maintaining it, delivering on it, facing "what next." Compared with the excitement of starting, the weight after finishing is far more real. So your body makes the call for you — fleeing right before the paper is due. You think you're lazy; really you're afraid that once the paper is submitted, life loses its excuse to stay in the easy "still preparing" zone.

You've trained yourself to be "good only at beginnings"

Every time you pull back at the decisive moment, your brain quietly notes: starts are fun, finishes are dangerous. So you get better at igniting and worse at landing the fire. A new project lights you up; the moment you hit deep water, you find a reason to leave. Eventually you accumulate a pile of "almost made it" half-things that serve as evidence for "I guess I can't do big things" — yet what's really stopping you was never ability, but the habit of braking right before the finish.

Give that "almost there" moment some cushion

Next time you feel the urge to let go, don't move. Treat it as a signal, not an order: ah, I'm not afraid of this task, I'm afraid of the chain of responsibility once it actually works. Then try only the smallest next step in front of you — send the email, run the code, without demanding you figure out the next three years. Often if you just survive that one beat, the thing lands, and landing is nothing like the horror your mind invented.

You're not unable to finish. You just need a bit more breathing room before the finish line — not another turn to open a new round.