In science in the US, a large fraction of the available jobs are temporary, and typically scientists take on temporary jobs (usually called postdocs) earlier in their careers. Why is this the case? Anecdotally, most occupations have a smaller fraction of temporary positions.
I'm interested in both answers about the historical events leading up to the status quo and answers justifying the status quo as a desirable situation.
Many of these positions are "soft money" positions - that is, they depend on grants. Since grants are temporary, the jobs they fund are also somewhat temporary. That said, this doesn't tell the whole story since temporary positions typically have a shorter term (for example, a year) than the grants that fund them.
For post-docs, it's a bit different. Post-docs are officially "training" positions - they are meant to be a step in the development of a scientist, a stepping stone towards more independence. Both employers (that is, universities) and funding agencies often have limits on how many years someone can be considered a "post-doc" because the idea is that this training should be temporary.
You could argue that the grad student to post-doc to professor track follows the apprentice/journeyman/master structure in the trades (I'm not certain whether it was explicitly inherited/motivated from that system, though).
Comparing academic to industry jobs in the US, while there are certainly differences in the hiring schemes I'm not sure they're actually all that different. "Permanent" jobs come to an end all the time as employers go through cycles of growth and layoffs. In most (probably all?) US states it is far easier to end someone's employment than it is in other countries like the UK or Germany.
It's a myth that "science jobs are temporary." The reality is that all jobs are temporary. Employers go bankrupt and employees die, but more frequently employees find a better job and switch. In the US, many employers are able to terminate employment at any time, which is less secure than having a one year contract with a fixed end date.
Many science jobs have a fixed duration of one to two years because many employers have one to two years of funding available. If the funding available increases, the job duration may be increased too.
The current situation is desirable from the point of view of awarding grants: If the funding is temporary, you can decide not to renew it when the awardee performs poorly. From the point of view of scientific staff, longer employment contracts would be better.
Short contracts will continue to be offered to scientists as long as scientists continue to accept them.
