The demo curse was a supposed phenomenon within the MegaZeux community which was cited as a cause of death for many potentially upcoming game releases--essentially, the idea was that if a demo was released, then it was far more likely that the project will be cancelled later down the line. As the practice of episodic and demo releases grew, it began to be considered favorable to release demos; however, as many of these demos were never completed, it appeared as if the act of releasing a demo had "killed" the project.
While the "demo curse" was typically cited now as a reason to avoid demos, however, the concept was a simple confusion of cause and effect. The games did not fail because a demo was released--rather, the games failed because of the demo mentality that caused its release in the first place. As demos became popular, eager developers would work toward completing a demo and releasing it as soon as there was enough content for one rather than focusing toward completing a game; satisfying the need for validation and attention that lead to the race for the demo release in the first place, the development pace would relax, and the developer would eventually lose interest in their project altogether.
It is still generally considered bad practice to release a game demo; however, it is important to acknowledge the actual reason why demo games fail to be completed and adjust your game development cycle to your own ability to hold attention to a project. Short games such as Lochronotran and Forgotten have been met with acclaim; it is better to start small and finish a game than to overaim and be left with nothing at all, or worse, a demo.
In the absence of demos, several people now leave "project dumps" with what they've done of a game; however, unlike with a demo, the intentions of releasing a project dump are a near-guarantee that there will be no full version of whatever's released.