So there is a lot of good but don't forget your question: haven't we had this for a long time? Yes, the techniques are fundamentally the same as others which have been used for a long time. Endonuclease and homologous repair are standard tools in genome engineering. It just costs much less to design custom endonuclease now. It seems like there is a bit too much hype about CRISPR/Cas9 techniques as genome engineering tools--- we are engineering genomes in exactly the same way as before. The scissors have changed but the glue is still endogenous to the organisms that we are engineering.
To my knowledge there has only been one case in which DNA was shipped as code to be the genome of a dead cell. Maybe someday we will be able to write large genomes. Until then nearly all the editing we do will be in living organisms, as it has been forever (even before CRISPR/Cas9).
Then CRISPR came along. It was like the open-sourced, better-performing alternative to the cumbersome, proprietary software. Switching was a no-brainer, and it has handily become the future, if not the mainstream already.
[1]https://en.wikipedia.org/wiki/Transcription_activator-like_e...
Previously, transgenics was tough mostly because of the difficulty in inserting sequence in the right place doing the right thing. It would take decades to develop the specific organism-specific tools to really change DNA.
>it's gene editing in an already living organism.
True. The thing about older transgenics, is that it was like using a shotgun to build a birdhouse. It was messy and you broke a lot of things to get the one gene where you wanted it. So you had to do a lot of transformations to get it right, and afterwords there was a lot of genetic cleaning up to do.
With CRISPR you can be extremely specific with your targeting, which means you can do things like "gene therapy", which is "patching" DNA in a live organism.
As for updating "live" code...that's a flawed, but not completely wrong analogy. Other techniques do rely on modifying the genome before "production", and in that sense CRISPR does enable us to edit DNA in cases that would have been impractical before, but it still basically requires performing the modification on each cell individually -- so there are still practical limits on deploying the technique.
https://en.wikipedia.org/wiki/Zinc_finger_nuclease
https://en.wikipedia.org/wiki/Transcription_activator-like_e...
For 'higher eukarya' the big problem is you can still do the long homology ends, but a competing process is random insertion. Basically (if my understanding is correct), CRISPR reduces the competing process and makes specific insertion of DNA the dominant result. Sometimes, though, having multiple random insertion is not a huge problem.
One thing that stands out to me (especially from the radiolab episode) is that it sounds like CRISPR isn't just gene editing in the sense of engineering something in a lab; it's gene editing in an already living organism. If DNA is anything like an organism's "source code", once the code is "shipped" (organism is conceived), traditionally we tend to think of that code as being locked/frozen. It sounds like CRISPR is akin to modifying the code live - "in production", so to speak. Is that a fair analogy?
Edit: to explain, when I say "in an already living organism", I'm referring mostly to a developed, multi-celled organism. I understand that traditional techniques also use living cells, but the radiolab episode makes it sound as if a full-grown adult human may someday get a live "DNA upgrade" - at least to applicable portions of the body - via CRISPR, e.g. to remove a genetic predisposition for developing a particular disease. To me, that would be substantially different (in practical application) from genetically engineering something like a gamete or a single-celled bacteria.