By the phase in the Pacific War where the Japanese began using the tokkōtai (特攻隊) or "Special Attack Units" in organised operations, the situation for Japan's once mighty air forces had changed dramatically. The Japanese had entered the war in 1941 with what was arguably the best naval air force in the world, capable of launching massed attacks from up to six flight decks at once, a capability the Americans would not acquire until much later in the war. That naval air force had been whittled away by casualties and attrition over several years of war (especially during the meatgrinder of the Solomon Islands campaign), and the nature of the Japanese pilot training system as well as substantial resource limitations meant that the quality of Japanese naval air power had fallen substantially. In 1944, at the Battle of the Philippine Sea, the IJN's carrier air groups, which they had spent nearly a year reconstituting after the battering they'd received in the Solomon Islands were completely smashed by the American carrier fleet, with the Japanese losing hundreds of aircraft with only a single hit on an American battleship to tell for it. The Japanese carrier air arm would never recover from the battering it received at the Philippine Sea, and it had become clear that conventional naval air battle tactics were ineffective. The pilots trained to replace those lost were simply not of a quality capable of pressing home an attack with any chance of success against the significantly improved American fighters and air defenses.

So what to do? With the losses suffered in the Philippine Sea, it was clear that Japan would never be able to reconstitute a naval air arm that could engage the Americans in a conventional manner with any hope of success. However, it was substantially easier for a pilot to fly their plane into a target (effectively acting as a guidance system for a missile) than it was for that same pilot to press home an attack with bombs or torpedoes. This was the bargain the Japanese took. They could not hope to put together enough skilled pilots for their air forces to be the same kind of threat they were at the beginning of the war, but they could put together pilots who at least knew how to take off, be guided towards a target area, and then pilot into a target. After all, most of the pilots being set to strike Allied fleets were being lost regardless of whether they tried to use conventional attacks or suicide attacks, but the suicide attacks at least resulted in more damage to the enemy.

In the long run, perhaps it was an inefficient use of pilots and aircraft, but for Imperial Japan in 1945, there were few better options. Japan's industrial base and economy were unable to produce or support the forces necessary to engage the Americans in a conventional manner. If they could, they undoubtedly would have continued to do, as they had from 1941 through to 1944. The use of suicide attacks was a desperation measure, borne out of the fact that the Japanese simply could not reliably use other tactics to any degree of success. Whether it was efficient or the best idea in the long term was effectively irrelevant. It was what was available.

There are three levels on which to consider the answer:

1. Were kamikaze attacks more effective than conventional attacks?

2. Did kamikaze attacks inflict more losses on enemy forces than they caused Japanese losses?

3. We can also ask the same question again, but consider the ratio of industrial capability.

The Battle of the Philippine Sea, in June 1944, demonstrated the poor return from conventional air attacks - the air component of the battle is nicknamed the "Great Marianas Turkey Shoot". On that day, the Japanese carriers launched 373 sorties to attack the US fleet. 243 of those were shot down. Some of those that made it back to their carriers were unserviceable. In their first wave of 69 aircraft, none reached the US carriers; some attacked some picket destroyers and some attacked US battleships, with a hit on South Dakota (which didn't stop her performing her duties). Of the 2nd wave of 129 aircraft, 6 reached the US ships and attacked (causing some minor damage, and some casualties, to Wasp and Bunker Hill through near misses). US losses in stopping those attacks were 17 aircraft (15 of those in combat). In summary, the number of Japanese pilots killed (not even counting other aircrew, or wounded) exceeded US casualties from the Japanese attacks, including aircrew and crew casualties on ships hit (or near-missed). Japan suffered further losses to aircraft and ships from the US counter-strike and from submarines, but those aren't relevant to the question of kamikaze vs regular attacks.

The pattern with the invasion of the Philippines in October 1944 was similar. Japan had learned from the Battle of the Philippine Sea and used their carrier force as bait rather than as a striking force, depending on land-based aviation and surface ships to attack the invasion fleet. The bait worked, but US air attack and US battleships took a heavy toll on the Japanese ships, and the Japanese surface ships that did get in close failed to take real advantage of their opportunity. US air raids on Japanese airfields that could be used to attack the invasion fleet kept Japanese forces far to weak to have any significant chance of achieving anything through conventional attack. Thus, the resort to suicide attacks. The first two attempts failed (probably). The first was 6 kamikazes, which were destroyed on the ground by a US air raid before being able to take off. The second was quickly put together as a replacement, but only consisted of 2 kamikazes and 1 escorting fighter. One of the kamikazes disappeared (but might have found and hit HMAS Sydney, which had a Zero crash into it on that day, which wasn't recognised as a deliberate suicide attack), and the other 2 planes returned. Attempt 3 was the first "large" attack: 6 kamikazes attacked a US escort carrier group, hitting two carriers. The same day, another 5 kamikazes attacked, again hitting two escort carriers. Result: 1 escort carrier sunk, 3 damaged by hits, and 2 more damaged by near misses. US dead on the ships exceeded 250.

The kamikaze attacks from the Philippines continued into January 1945, by which time about 650 kamikaze sorties had flown. Japanese losses were greater than 650 aircraft, since kamikaze missions were often escorted by regular fighters. In total, this was about double the air effort of the Great Marianas Turkey Shoot, spread out over 3 months instead of concentrated in one day. Of these kamikazes, 364 reached US (or other Allied) ships. 231 were shot down by anti-aircraft fire, and 113 hit ships, and 56 caused damage from near misses. Conventional air attacks continued over this time, too. In summary,

Type	Kamikaze	Conventional
Sorties	650	?
Reached ships	364	1092
Shot down by AA	231	156
Hit	113	23
Damaging miss	56	?

First, it can be seen that far more conventional attacks got through US air defences than in the Battle of the Philippine Sea - US defences were much lighter over those 3 months. Second, kamikazes were (in principle) not deterred by AA fire, and would continue to attack into AA fire that would often deter conventional attackers.

Third, not apparent from this data, is that the attacks were most effective on the first few days. This opening phase resulted in 47 ships hit (including 7 escort carriers), of which 5 were sunk and 23 heavily damaged. This was achieved by fewer than 100 sorties. A small effort compared to the Battle of the Philippines Sea, and with fewer losses (despite these being kamikaze attacks), brought much greater return. The USN learned, and became much more adept at stopping kamikazes. Over the whole 3 months, a total of about 20 ships were sunk and about 110 heavily damaged by kamikazes.

The US took the lessons learned from the kamikaze attacks in the Philippines with them to Okinawa, e.g., using radar picket ships to obtain earlier warning by radar (kamikazes usually flew low to avoid detection, and could get quite close before being detected by radar). One Japanese innovation at Okinawa was large-scale kamikaze attacks; the first of these (and the largest) consisted of 355 kamikazes and 340 other planes (escorting fighters and bombers making conventional attacks); 33 ship were hit, 4 of which sank. In total, 1500 kamikazes flew in such large attacks, and 400 in smaller attacks, in the fighting for Okinawa. In total, off Okinawa, 26 ships were sunk by kamikazes, and 225 damaged. US casualties from kamikaze attacks were probably about 8,000 (about 4,000 killed and 4,000 wounded). 10 ships were sunk and 143 damaged by other causes. Japan lost about 7,000 aircraft in the battle; the kamikazes were less than 1/3 of this total, and achieved the majority of the effect.

So, to return to my two initial questions:

1. Were kamikaze attacks more effective than conventional attacks?

A: Yes.

2. Did kamikaze attacks inflict more losses on enemy forces than they caused Japanese losses?

A: In terms of casualties, yes, but not overwhelmingly so.

In both the Philippines and Okinawa, Japan made a "maximum effort" kamikaze effort for the duration of the battles. Attacks were halted in order to build up forces for the next attack. Just as the kamikaze effort in Okinawa was larger than in the Philippines, the kamikaze effort to stop an invasion of Japan would have been even greater. While losses would probably have been higher than at Okinawa, there is no reason to think that the invasion would have been stopped.

In summary,

• Kamikaze attacks were significantly more effective than conventional attacks, in terms of damage inflicted compared to losses. Kamikaze pilots received less training than conventional pilots (about 1/3 or 1/2 of the flight hours), so kamikaze losses were more affordable by the Japanese pilot training system.

• They were not effective enough to achieve their goals. They neither stopped the invasions they tried to stop, nor inflicted militarily unaffordable losses on the US and Allied forces.

To return to your initial question:

Wasn’t it a bad strategy in the long run?

Of the main choices available: (a) continue with purely conventional attacks, (b) use kamikaze attacks, (c) stop attacking invasion fleets, and (d) surrender, it was a better strategy than (a) and (c).