His main arguments are these:
The Japanese placed immense faith in their kamikazes. They hopes to fling thousands of kamikaze planes at the invasion fleet en masse, in hopes of overwhelming the US anti-air network and causing massive damage to the invasion fleet. Sounds great in theory, but would not have worked in practice. In order to continue to protect those thousands of planes, they had to disperse them all across the home islands and almost literally bury them in camoflage to prevent them from being destroyed on the ground. First, the virtual absence of radios and breakdown of communications on the home islands would have made it impossible to effectively coordinate and mass the planes before the strike. Also, as soon as the planes began to mass, they would be destroyed. There would likely have been piecemeal commitment of kamikaze aircraft in groups easily handled by what would have been a massive air-defense network.
There were two operations planned: Operation OLYMPIC, which was to seize the southern third of Kyushu and turn it into a massive air and naval base, to be used in support of the next operation, CORONET. CORONET's objective would be to invade the Tokyo Bay area, and seize the Kanto plain, the industrial and political heart of modern Japan.
The Japanese decided their best hope was to meet the invaders at the beach, hold them there with Okinawa-style fixed defenses, and drive them back into the sea with a massive counterattack. The idea was to create a bloody free-for-all melee on the beaches so American air and sea power would be unable to intervene. They really did not expect to succeed, but hoped to make the affair so bloody as to force the Americans to seek a negotiated peace.
But, while this attritional strategy had worked on Luzon, Iwo Jima, and Okinawa, it would have been much harder to carry out on the home islands. While the previous battles had been fought with veteran troops, the homeland defense units were extrememly poorly trained, and in most cases their equipment was woefully inadequate. Also, by the time the invasion was to occur, the Okinawa-style fixed defenses would have been nowhere near ready.
Their massive counterattack would also be near-impossible to effectively carry off. Allied air power would make strategic movement impossible, and would make tactical movement extremely difficult. As a result, their counterattack forces in all locations would have been committed piecemeal and easily defeated.
Claims that the Japanese planned to mobilize the entire civilian population and use them as a last suicidal defense are totally unfounded. In reality, the civilian mobilizations were designed to let them handle rear-area work so more soldiers could be freed up for combat. In fact, they even made plans to evacuate civilians from combat areas. One Japanese commander remarked that the idea of using civilians as cannon-fodder drastically hurt morale in his unit. Besides, even if it had been done, it is doubtful they would have had much effect. Untrained women, children, and old men wielding sticks would not be of much use on a seasoned enemy armed with armor, artillery, rifles, submachine guns, and flamethrowers.
The Japanese staked everything on defending against OLYMPIC. Once it had succeeded, it is likely they would have surrendered and CORONET would not have been necessary. Even if it had, the Japanese had never seriously thought much beyond defending Kyushu, and thus would have been horribly unprepared. Also, the terrain of the Kanto Plain decisively favors American mobility and totally cancels out the Japanese fixed-defense strategy. With the seizure of the political and industrial heart of Japan, as well as Tokyo itself, the war would have certainly ended.
The claims of "one million casualties" have no basis in reality. The actual estimates were closer to Okinawa and Normandy. It is likely that these were accurate.