A man like Mikhail Tukhachevsky needs no introduction, as he is one of the most famous military minds of the 20th century, having come up with the very effective Deep Battle doctrine. However, some of his ideas were rather odd, even for their time.
"In my note on tanks, I wrote: "If we take Magdesiev's calculations as 50%, then with our plans for tractor production in 1932/33 of 197,000 units, the annual tank production can be 100,000. If we consider one year of war to consume 100% of tanks (this is an arbitrary number), then we can maintain a park of 50,000 tanks."
Why 50,000 tanks from 100,000? Because in my proposal, the building of 50,000 tanks will be done not in peacetime, not during mobilization, but in the first year of war. So out of 100,000 tanks , 50,000 (excluding tanks build during peacetime) will be spent on deploying the tank units and 50,000 will be used to compensate for 100% losses. I don't understand how people who are well informed in issues of military organization, and Staff workers must be so, can't figure out such an elementary issue.
The RKKA Staff considers the loss of tanks per year of war to be 400%. In this case, we can maintain a fleet of 20-25,000 tanks. Characteristically, the RKKA Staff did not apply their standard to my note, but the opposite, took 50,000 tanks as the starting point, applied their 400%, and calculated that a peacetime order of 142,000 tanks must be made. I can't find the words for this kind of behaviour in the Staff.
According to my calculations, 8-12 thousand tanks must be mobilized to organize a new kind of deep battle, which I told the Revolutionary Military Council having not yet read your letter. I know there is nothing fantastical about this number, which can seen from such things like comrade Voroshilov stating that the government offered us 10,000 tanks, but he thought it was unreasonable and proposed 3,500 tanks.
Let me add that we still treat the tank issue very conservatively, insisting that all tanks be of the military type. Considering that we will have no competitors in Europe in the field of tractor building by the end of the five year plan, we need to strive for having 1/3rd of all tanks be special military types, to perform special tasks of fighting enemy artillery, etc. The rest of the tanks, in the second and third echelon, can be slower, larger, etc. This means that such a tank can be an armoured tractor, just like we have armoured cars, trains, draisines, etc. We can produce massive amounts of armoured tractors.
...
The RKKA Staff insists that it is necessary to build many large military factories, which I consider incorrect. Military production can be based on civilian factories, which I proved with figures attached to my notes on industrial mobilization and the artillery program. The attached notes explain that I aim to minimize peacetime spending by way of adapting civilian products and organizations to the aims of war."
Even though Tukhachevsky fell out of favour and was eventually executed, his tractor tank idea was tried during the desperate early months of the Great Patriotic War. These tanks were of limited usefulness, and did not last long in production, making room for much more useful "second echelon" tanks.