Is Cockshott legit?

Yeah. He has his critics, but his work on LTV and economic planning are worth engaging with. However, he is a British terf, so I hate that the above is true.

Cockshott clowned on this guy

Some brief background on the labor theory of value (LTV):

The LTV says that the price of a product mainly reflects the cost of the labor that went into it. For example, you buy a toaster. Workers had to assemble the toaster, drive it to the store, mine the iron to make the metal parts, drill the petroleum to make the plastic parts (and fuel the trucks, and power the electric grid), and so on, across the supply chain. If you follow this logic out, the production cost is labor all the way down.

Because companies compete to sell the cheapest toaster, the price of a toaster will be close to the cost of making the toaster. And because that cost is mostly labor cost, the price therefore reflects the labor that goes into the product. If you want to charge much more than that, you'd better work on your branding and/or find ways to make your toaster unique so it has its own little market niche where there's less competition.

You might be thinking, "wait, doesn't price come from supply and demand?" Not primarily. Supply and demand are a red herring. On short time scales, if there's a toaster shortage, the toaster companies might price gouge a bit, but eventually the market adjusts, demand goes back down or production rises. As long as there's sufficient competition, prices ultimately fluctuate around an equilibrium that is pegged to production cost.

The name of the game, in consumer capitalism—after you suppress wages as much as possible—is to get around competition so you can charge more. Hype up your brand image. Make your product unique. Sell in places others don't sell. Stores have endless variety because variety means no one exactly competes with anyone else and everyone can charge a little more. Up to a point.

Summary of Cockshott's analysis defending the LTV, as I understand it

What Cockshott did was add up everything that went into a bunch of products: all the labor that went in, all the oil, all the electricity, all the wood, etc. Note that this was across the entire supply chain. That means there's inherent redundancy: labor includes labor to drill oil, produce electricity, and chop wood, for example. Labor input and oil input are not independent variables. This will become important in a second.

Cockshott found that price correlated with labor more strongly than with any other input.

One of many reasons why the medium article is full of shit

Remember how labor input and oil input are not independent variables?

Here's what the Medium guy says:

This [Cockshott's analysis] is quite an unusual method: typically when statisticians want to test one cause against another, they don’t correlate them one by one. Instead, they use multiple regression analysis or some other method that includes all variables at once. Cockshott’s approach is a bit like if those who studied wages and education correlated the two, then separately correlated wages with gender, then age, then parental education, and concluded from the largest correlation that said variable was the key determinant of wages compared to the others. Spot the problem?

Multiple regression only works for independent variables. Production inputs are not independent variables, everything is related to everything else. A supply chain that uses more labor probably also uses more oil and electricity. For that matter, the Medium guy even fucks up his own example: education and parental education are correlated.

Cockshott says as much here, if I understand correctly:

He [Medium guy] critiques us for looking at the correlation between price and electricity steel etcetera content separately saying we should have measured them at the same time using a multiple regression. He seems to have again overlooked the fact that when we’re looking at steel content, when we’re looking at oil content or labour content we have in all cases followed back the production processes in terms of all the inputs, we decompose them into either their labour content, their steel content or their electricity content etc. So, when we give these different correlations, we are giving the correlations considering all the inputs so can’t sensibly do a multi factorial regression here when you have already taken them into account in the respective Leontief inverses in each case.

I googled what a Leontief inverse is, and it's just a big matrix saying how much of which stuff went into which other stuff.

This comment is long enough so I'll stop here and go to bed.