To claim "no experience" overstates to the point of misleading, in the case of ordinary breeding. What we get from some of the GM stuff is new kinds of genetic combinations - not minor variations within a familiar and well-experienced framework. The problem is not novelty, in a trivial sense, but a degree of novelty so large as to be qualitatively different - to be unpredictable to a much larger degree and in completely unfamiliar ways. This is false. No commercial GM technique alters only one gene, the change in the genome is not usually that tightly controlled, the total effects of the new genetics are difficult to track and impossible to fully predict (in part because the workings of the host genome are not well enough understood, in part because all the possible effects of the new gene are not known and cannot be reliably observed unless predicted and watched for), and so forth, and so the kinds of effects, not just the range but the kinds, possible are not known - unlike with conventional breeding, where the new string was formerly in a very similar environment and could be observed there, the new genetics here are in an environment much different from their old one and never before observed reacting to them. Some risk of genetic uniformity is imposed by inserting the same genetics into every modified organism - if a lethal and virulent pathogen ever learns to exploit the specific framing and insertion genetics of Round-up Ready soybeans, for example, we lose most of the US soybean crop at one stroke. The genetic uniformity is also increased by the circumstance that each modified variety is another expense, and so cost considerations limit the genetic variety of the modified crops. In responsible use, as with responsible use of antibiotics, applications can be timed and limited to avoid the kinds of unremitting sublethal exposure that engender resistance quickly (as described by standard Darwinian theory). In GM crops, no such regulation is possible. If you were to set out to breed resistance, informed by Darwinian theory, you could hardly do better. That is what scares me, about the tech folks - they do not recognize when they are walking off a clliff. These genes are in brand new environments, and they were often put there in ways that abet them jumping around and changing further. They are like an organism landing on a new continent and given extra mobility - the possibilities are orders of magnitude more varied than the those of the adjusted and mutually adjusting natives. My intent was not to dismiss the potential of the GM salmon, but to illustrate the enormity of the hazards of the more risky forms of GM. The threat with the salmon is predictable, trackable, measurable, known, sharply limited to salmon initially; compare with the threats from jumping herbicide sequestering bacterial genes framed in auxiliary abetters from unrelated plants and inserted into insect and fungus connected beans and grasses planted across a continent, say - for human and animal consumption. Hello? But they aren't nearly as poorly characterized - they are all pre-fit into a system whose product we have long experience with. Their shuffling is limited, their expression subject to pre-evolved limits and feedbacks and curbs within the organism. They are expressed within an environment pre-adjusted to handle them. They are not set up to be transferred easily into viruses, or moved between different organisms. They have a role, and we are familiar with their playing of it. Any of dozens, but an example you can check easily would be the sequestering of herbicide through chemical binding in plant organelles, the mechanism of resistance in glyphosphate-resistant crops. It turns out that some of this gets into the food produced (unexpectedly, it was supposed to stay in the leaves) and it turns out that when digested in a person's small intestine full of bacteria it sometimes happens that the various chemical constituents of the herbicide complex are released - modified, whole, broken down, etc - into the person's body. And nobody as of right now knows whether this is trouble or not. We're too committed to the poorly studied and dangerous GM crops - when something like this turns up, we can't even put a moratorium on the planting and harvesting until we figure out whether we are in a disaster.. In a real world ecosystem, that would take a century or more. No matter, no one is doing it: too expensive, too complex. The possibilities one would have to check are orders of magnitude greater than for conventionally bred crops. The tool is no better than Monsanto's intentions, or Cargill's, or Dekalb's. And possibly worse.