Your objection has some validity, up to a point. What changes is average cost of the inventory items involved, not overall inventory value. What was described is a straight barter transaction, where the value of one commodity or lot has been agreed to equal the value of the other. So the values of both subsidiary ledgers in Inventory on hand remain unchanged if you take the quantities-only approach. But since the quantities change, the average costs change.
Imagine an inventory holding 5 of Item A at $1 per unit average cost ($5 total in Inventory on hand) and 10 of Item B at $2 per unit average cost ($20 total in Inventory on hand). You barter 2 of Item A for 1 of Item B, with no money changing hands. Item A now reports 3 units with total value of the same $5, for a new average cost of $1.67. Item B shows 11 units at the original value of $20, for a new average cost of $1.82.
Now consider a situation that starts with the same conditions, but uses separate payments and receipts instead of a barter transaction. You sell 2 of Item A for $2 (zero markup to keep things simple). You now have 3 units on hand worth $3 total. You buy 1 of Item B for $2. You now have 11 units on hand worth $22. Average costs for both inventory items are unchanged at $1 for Item A and $2 for Item B.
Notice that the grand total of Inventory on hand for both scenarios remains at $25. Why the difference between resulting average costs after what seem on the surface to be equivalent exchanges? The barter transaction (the first scenario) effectively fixes the purchase cost of 1 additional unit of Item B as the sales cost of 2 units of Item A. But the receipt-and-payment scenario actually transfers monetary value between the subsidiary ledgers for Item A and Item B.
So yes, in this simplistic example, I concede the receipt-and-payment approach would be more accurate. But the effect diminishes both with higher quantities on hand and larger turnover volumes. Since the overall value of inventory is unaffected, it may not be worth the effort.
I believe what you are advocating is that you should include monetary values in the barter transaction, presumably to achieve the same result as the receipt-and-payment approach. I agree that would technically be more accurate at fine scale. To do that, though, you also have to consider the need to balance the journal entry by which the barter transaction is entered.
If you credit Item A $2 (-2 units x $1 average cost), representing your cost of goods, you would need to debit Item B by the same amount. The result is exactly the same as my first illustration—no changes to average cost. But that is only because I deliberately chose values to make things come out evenly. What if the current average cost of Item B was $2.10? Then, to balance the journal entry, you would enter the debit at $2 for 1 unit, and the average cost would be lowered. What makes that new average cost more valid than the old one, since it is unrelated to any purchase of Item B, only to purchases of Item A?
To flip things around, you could also record a journal entry where the cost deducted from Item A was determined by the current average cost of the unit of Item B being acquired. But is that any more accurate? And what if you have none of Item B on hand, and therefore cannot determine an average cost?
I think this reflects why tax law and accounting standards surrounding barter exchanges are frequently complex.