The Falcon 1 had failed three times before being successful, Walter Isaacson explains (in his biography of Elon), and the Falcon 9 was far bigger and more complex:
The chances for success were not helped when a storm rolled in and soaked the rocket. “Our antenna got wet,” Buzza recalls, “and we weren’t getting a good telemetry signal.” They lowered the rocket from the launchpad, and Musk came out with Buzza to inspect the damage. Bülent Altan, the goulash-cooking hero of Kwaj, climbed a ladder, looked at the antennas, and confirmed that they were too wet to work. A typical SpaceX fix was improvised: they fetched a hair dryer, and Altan waved it over the antennas until the moisture was gone. “You think it is good enough to fly tomorrow?” Musk asked him. Altan replied, “It should do the trick.” Musk stared at him silently for a while, assessing him and his answer, then said, “Okay, let’s do it.”
The next morning, the radio frequency checks were still not perfect. “It wasn’t the right sort of pattern,” Buzza says. So he told Musk there might be another delay. Musk looked at the data. As usual, he was willing to tolerate more risk than others. “It’s good enough,” he said. “Let’s launch.” Buzza assented. “The important thing with Elon,” he says, “is that if you told him the risks and showed him the engineering data, he would make a quick assessment and let the responsibility shift from your shoulders to his.”
[…]
The day before the planned December launch, a final pad inspection revealed two small cracks in the engine skirt of the rocket’s second stage. “Everyone at NASA assumed we’d be standing down from the launch for a few weeks,” says Garver. “The usual plan would have been to replace the entire engine.”
“What if we just cut the skirt?” Musk asked his team. “Like, literally cut around it?” In other words, why not just trim off a tiny bit of the bottom that had the two cracks? The shorter skirt would mean the engine would have slightly less thrust, one engineer warned, but Musk calculated that there would still be enough to do the mission. It took less than an hour to make the decision. Using a big pair of shears, the skirt was trimmed, and the rocket launched on its critical mission the next day, as planned. “NASA couldn’t do anything but accept SpaceX’s decisions and watch in disbelief,” Garver recalls.
[…]
SpaceX repeatedly proved that it could be nimbler than NASA. One example came during a mission to the Space Station in March 2013, when one of the valves in the engine of the Dragon capsule stuck shut. The SpaceX team started scrambling to figure out how to abort the mission and return the capsule safely before it crashed. Then they came up with a risky idea. Perhaps they could build up the pressure in front of the valve to a very high level. Then if they suddenly released the pressure, it might cause the valve to burp open. “It’s like the spacecraft equivalent of the Heimlich maneuver,” Musk later told the Washington Post’s Christian Davenport.
The top two NASA officials in the control room stood back and watched as the young SpaceX engineers hatched the plan. One of SpaceX’s software engineers churned out the code that would instruct the capsule to build up pressure, and they transmitted it as if it were a software update for a Tesla car.
Boom, pop. It worked.