A "good" modification would require:
No patents
No suing farmers that grow your seed without a license/purchase
No boosted tolerance to plant poison
No convenient requirement that "improvements" rely on additional product that the company sells (eg roundup)
Edit: formatting
People won't buy such things for a long time anyway, and that it doesn't occur in nature doesn't matter. Hell, give it a million year and it might start occuring in nature by itself, because why not? Nature does what nature does, it already mixed up a lot of stuff in a weird way (except we don't call them "weird", just "natural").
Few other things that "never happen naturally" - clothes on human beings. Bricks. Steel. Airplanes. Colorful paint on canvas. Should we discourage them all, because they're not "natural"?
Compare this with nuclear weapons - humans have in a few decated managed to engineer something way more dangerous than millions of years of "natural" permutations/combinations of atoms.
This is 100% False. There is essentially no risk of accidentally transferring a bacterial protein to a plant. The molecular machinery involved in transfection is complex and fragile. The transfer process is gated and specific. Even if there was a transfer of the wrong gene the result would be a dead organism, not a super-organism.
Eukaryotes do not tolerate random prokaryotic genes in their genome in place of a functioning gene. The introduced gene has to be specifically designed/selected to be compatible with the target organism.
Whoever is telling you this is doing the exact fear mongering these Nobel laureates are advising against.
It is so much the opposite of easy to do what you describe that it is essentially impossible.
GMOs should certainly be regulated for safety and transparency, which is part of the job of the FDA in the United States.
Dekhn's statement in the comment about the allergenicity was this:
A gene from Brazil Nuts that coded for a protein high in methionine was cloned into soy plants; soy from the resulting plant caused allergic reactions in subjects.
An important missing part of that ...in subjects with an allergy to brazil nuts. In otherwords they found some allergic reaction to the brazil nut protein in people allergic to brazil nuts. It did not create an unknown or novel allergen (although allergenicity should be a standard test for GMO food products).
For 8/9 subjects it was not the primary allergen (and for one it was the allergen that causes a dangerous reaction). Regardless an allergen is present and it is good that it was identified. The paper concludes with this:
"It is prudent to assess the allergenicity of proteins in transgenic foods if those proteins have been derived from sources that are commonly allergenic. The use of currently available animal models alone to predict allergenicity in humans does not produce accurate results. Techniques such as radioallergosorbent tests, SDS-PAGE with immunoblotting, and skin-prick testing can identify IgE-binding proteins that are probable allergens in transgenic foods derived from sources known to be allergenic. This strategy will not be useful in assessing the allergenicity of transgenic foods in which the allergenicity of the source of donor genetic material is unknown. Most current applications in the field of plant biotechnology fall into the latter category."
There is no doubt (in my mind) that GMO needs to be regulated and reported. If there is any protein from an allergic source that source needs to be listed on the food product specifically as "contains material from X" -- in this case X = brazil nuts.
The fact that the source donor material could be unknown (proprietary secret) is completely unacceptable. The FDA should have full regulatory control over GMO and they should be treated with a scrutiny closer to drugs rather than food. There should be absolutely no "proprietary secrets" regarding the source material or process used to make the modifications. There should be complete transparency and rigorous testing for allergic potential.
Maybe increased regulatory control is all that GMO doomsayers are after, but I think the hyperbole hurts their argument when there are sound scientific arguments for increased transparency and regulation.
tl;dr: This is a legitimate concern that should be included in a regulatory framework that supports safety and transparency.
There is still a small chance that someone who did not know they were allergic to Brazil Nuts but already consumes soy products could be impacted by the new product. However I am not sure how to mitigate that and still have GMO food products. From a consumer perspective they could be introduced to allergens that are now present in their "new" existing diet. I would sympathize with a consumer who said "hey I did not change anything and now I am impacted." So in this case labelling would be insufficient. Perhaps the answer is to say that "known allergenic introductions" are not allowed for GMO food.
Many benefits and concerns lie ahead. In such cases, for their sake, the consumer needs access to valuable information.
Please consider that the vast majority of scientists who work in this area disagree with your logic - that is what the entire NYT piece is about.
http://www.npr.org/sections/goatsandsoda/2015/05/05/40419855...
What we know is what has happened in "nature", and while we can try to extrapolate from that, we really have no clue as to what is possible.
Even though you grossly exaggerate how genes work with regards to resistance/danger, the possibility of phenotypes that could lead to negative effects in human consumers is just as real by "natural" means.
I don't really think your comparison with nuclear weapons works though; for one, there is no such thing as a "natural" nuclear weapon - not to mention that a "natural" (objective) concept of "weapon" (irrespective of any subject making/wielding/using/thinking about/whatever) would have to exist if we were to compare apples to apples.
The idea behind my comparison is simply, that we can engineer things much more dangerous than nature can, that would never appear in nature by themselves.
But, generally, that's what we do. With brains, we can do in a few months what would take natural selection millions of years to "discover". Doing that, using our brains, seems kind of like the whole point of being a human.
1) You can't patent seeds unless you have modified them genetically. Occurrences of nature are not patentable. No GMO, no associated business practice.
2) Some of the modifications (I'm looking at you Monsanto roundup ready) allow plants to be tolerant of increased quantities of herbicide. This is obviously not something you want to encourage on a global industrial scale. Like you said, we have the food, we don't have the distribution.
3) That said, there is absolutely no inherent danger of genetic modification. We do it all the time, it's well controlled for effectively getting a specific modification. Calling GMO "frankenfoods" or implying there is some inherent health risk to GMO food (outside specific problematic modifications) is just bullshit.