The astonishing gains in agricultural output since the 1960's has come at an environmental cost. It has also failed to bring much benefit to sub-Saharan Africa, the region of the world that is predicted to have the greatest population increase in the coming decades and greatest agricultural need. The increase in productivity required to feed the billions that will soon be added to the world's population is more likely to succeed if creative direction lies in the hands of those most intimately acquainted with local agricultural concerns. Nitrogen has been a major limitation, particularly in sub-Saharan agriculture, and I look here at the prospect for extending biological nitrogen fixation to the crops and practices of local interest. A source of optimism in this historically intractable problem is the discovery in certain algae of natural organelles – nitroplasts – derived from unicellular cyanobacteria that fix N2. Transferring nitroplasts to land plants seems possible only if the organelles rely little on their hosts. A survey of the metabolic capabilities of three of them indicate different degrees of independence. The nitroplast from the diatom Rhopalodia gibberula may need nothing from its host except a source of carbon (the nature of which is explored). The nitroplast from another diatom, Epithema turgida, is somewhat less independent, while the nitroplast from Braarudosphaera bigelowii (called UCYN-A) requires a great deal from its host, including imported protein. The evolutionary route to acquire proteins may have been greatly simplified if nitroplasts share proteins with chloroplasts, a co-resident cyanobacteria-derived organelle. The genomes of the diatom nitroplasts suggest the presence of a few enzymes that are rare in cyanobacteria but seemingly well suited to the special needs of a nitrogen-fixing organelle. I describe how the host-range of nitroplasts might be explored and how a stripped down version, called a neoplast, might simplify strain modification to such a degree that the process becomes accessible to academic researchers with limited resources, even to individual farmers.