[wmkn]

What does 'failure' mean in this statistic. Does failure always mean a catastrophic failure (i.e. the recent SpaceX launch) or are failures also rockets not leaving the platform?

I would think that a >5% failure rate is surprisingly high for something we have been doing more than 70 years.

[bunderbunder]

If you follow the handy link to the source that the article provided, you'll get some more information.

Incorrect orbits are counted as failures, and there are some additional tables that break down the failures by type. It does not look like the rocket simply not going up is counted in the statistics - I'm guessing that always represents an abort rather than a true launch attempt.

[manarth]

Incorrect orbit of a secondary payload is also included in the failure count (I was thinking of a 2012 Falcon 9 launch, where the primary payload was successful but the secondary payload didn't reach its intended orbit).

[oh_sigh]

I've never heard of a rocket not leaving the platform. Every launch I've seen, the rocket left the platform, or destroyed itself on the platform.

I think(but don't have the data on hand) a more common failure scenario is the rocket launching, and the payload getting into space, but not able to attain the desired orbit.

[imaginenore]

The recent SpaceX event was not a launch. It was a test a few days before the scheduled launch.

[Heliosmaster]

In the last 70 years we've focused a lot on not having failures, spending all the money it took.

Now, the focus is somewhere else (namely reducing the cost per launch, in order to get us a lot more to space).

[kiba]

Pretty sure that SpaceX is focused on reliability. One of the analogy that Elon Musk used is comparing a Honda Civic versus a Ferrari, asking which is more reliable.

Historically, rockets are focused on performance, not cost.

SpaceX's focus on producing cheaper rocket and launching more of them will yield more reliability. You make more of them, you make more mistakes, more opportunities to correct them.

Eventually you'll have a system for making very reliable rockets.

[rimantas]

there is Honda Civic, never heard of Toyota Civic :)

[kiba]

Fixed.

[pjscott]

Accepting lower reliability of rockets doesn't buy you as much cost reduction as you might expect. Payloads are usually very expensive, so even a moderately increased risk of losing one can be pricy.

What SpaceX is doing is accepting worse performance in exchange for lower cost. Hopefully that'll make things more reliable in the long run, rather than less, since they won't be pushing the envelope quite as much as competing designs. (Remember the Space Shuttle Main Engines, which had to be practically rebuilt after each flight? That was done in order to get the amazing performance needed to make the Shuttle work at all.)