It measures material production, which is a key feature of capitalist economies, centered on the accumulation of material wealth. It was not designed to measure everything. Certainly not sustainability. Or happiness, for that matter. And although Diane Coyle, author of the book, suggests that it can't be used as a measure of well-being, GDP per capita is certainly employed as an index of welfare, even though no serious (I don't mean mainstream) economist would take GDP per capita as the only indicator of development. Finally, it isn't a measure of inequality, but functional income distribution, the shares of labor and capital compensation in total GDP, is actually one of the best measures of inequality. So yes, GDP does have its flaws (for an accessible discussion of the limitations of GDP go here, and for my views here).
Note that Kuznets apparently was against including government spending, in particular defense expenditures, as part of GDP, according to Coyle, since in his view it didn't increase well-being. That proposition is not uncontroversial. A lot of government spending, including in defense, is central for technological innovation, and, hence, for higher productivity growth and increasing living standards. The internet (as well as driverless cars and many other things) that I'm using to post this piece is the result of DARPA's investment -- the Defense Advanced Research Projects Agency, a defense department agency.
This suggests to me that we owe more to the British Keynesians, Richard Stone and James Meade, that basically created the methodology of of the National Accounts during World-War-II, than to Kuznets (all three won the Sveriges Riksbank Prize, by the way, but Meade's wasn't related to national accounting). And, in a sense, it is what Coyle suggests when she argues that:
“It [Keynesian economics] became the basis for a more interventionist approach to government economic policy from the 1940s onward, using both fiscal policy (the level of tax and spending) and monetary policy (the level of interest rates and availability of credit) to target a higher and less volatile rate of growth for the economy. The use of these tools was developed more fully by other economists after Keynes’s early death in April 1946. Postwar policymakers still bore the scars of the Great Depression and pounced on the economic theories of Keynes and his successors as a means of averting a repetition of that crisis. Crucially, the development of GDP, and specifically its inclusion of government expenditure, winning out over Kuznets’s welfare-based approach made Keynesian macroeconomic theory the fundamental basis of how governments ran their economies in the postwar era. The conceptual measurement change enabled a significant change in the part governments were to play in the economy. GDP statistics and Keynesian macroeconomic policy were mutually reinforcing. The story of GDP since 1940 is also the story of macroeconomics. The availability of national accounts statistics made demand management seem not only feasible but also scientific.”This is essentially correct, yet it might be misinterpreted as suggesting that the National Income and Product Accounts (NIPA) are intrinsically Keynesian, as some far right supply-siders have argued. Nothing in the NIPA implies that causality goes from autonomous spending to income, as in Keynes' Principle of Effective Demand, and the accounts are compatible with a model based on Say's Law (not that I personally think that's a good idea).
It only means that by the time the National Accounts were developed, a version of Keynesian economics, as it turns the Neoclassical Synthesis, had more or less become the mainstream interpretation of how the macroeconomy works. The same is true of say econometrics, and macro econometric models, like the Klein-Goldberger, which was Keynesian, and would not lead a reasonable person to conclude that econometrics is Keynesian (Keynes was, in fact, skeptical about it).
