I've been flabbergasted by just how good Winterspeak has been lately. I don't agree with all of his conclusions, but the perspective he's developing is quite beautiful, and very useful. I'm going to highlight a few of the bits I really like, and add a bit of spin.
I want to start with a related insight from the excellent Mencius Moldbug. Apologies to Mencius if this isn't the best reference, but the point is very clearly stated, in a comment at Brad Setser's:
Money is the bubble that doesn’t need to pop. As long as there is demand for indirect exchange, at least one asset will be stockpiled by hoarders, hence experience demand that is not a consequence of any direct utility, hence be overvalued. As long as the storage cost for this asset is zero and the supply in existence is fixed, you have a perfect Nash equilibrium - using any other asset as a medium of indirect exchange provides no advantage, and runs the risk of buying into a bubble which will subsequently pop as punters revert back to the stable standard.
This idea stands in stark contrast to the view recently offered by John Cochrane:
The value of government debt, including money, is equal to the present value of the primary surpluses that the government will run in order to pay off the debt. Nominal debt is stock in the government, a claim to its taxing power.
I think Mencius' view is more accurate. 
The next insight I will attribute jointly to Mencius and Winterspeak (while quoting neither): When a government implicitly or explicitly guarantees deposits and other bank debt, it is useful and in some ways more accurate to describe the arrangement as loans by individuals to the government and then separate loans by the government to banks. If bank loans sour, it is the government who is out the money. Bank creditors do not discipline banks, or select banks based on their investing acumen. The connection between the deposit base of a bank and that institution's ability to lend is formal and vestigal, and is disappearing as reserve requirements become an ever less binding constraint on bank lending. (I'm pretty sure I've read both Winterspeak and Mencius make this point, but I can't find great quotes. Here's an example from Winterspeak.)
This idea foreshadows the new Winterspeakian synthesis that has me suddenly awestruck. Winterspeak presents a view of the monetary/finacial system in which the government is ultimately the hub. As above, when we "save" without explicitly investing, we don't lend to banks which then lend to businesses. We lend to the state — we hold instruments that are ultimately claims against the state — while the government organizes the distribution of loans, implicitly via how it structures bank lending or explicitly by various forms of directed credit. But the government's role is deeper than that. Here's Winterspeak's counterintuitive, but obvious-once-you-think-about-it view of taxation:
Everyone agrees, more or less, that the Federal Government has the power to print money. That's what fiat money means by definition. Since the Federal Government can print it's own money, in whatever quantity it chooses, have you ever wondered why it taxes at all? In fact, if you chose to mail in your 2009 taxes in crisp Federal Reserve Notes, the Government would take that paper money and shred it.
The Federal Government does not need taxes in order to spend. At the Federal level, because the Fed is a currency issuer, the sole purpose of taxes is to extinguish money, reduce aggregate supply, and therefore limit inflation to a tolerable level.
But if the government destroys money, it also creates it:
The Government creates money and transfers it to the private sector by spending. The private sector transfers the money amongst itself (spending, investment) and puts the rest in the bank. If the private sector wants to put more money in the bank, that money can either come at the expense of transactions, or it can come from the Government simply running a larger deficit, and thus creating additional money for the private sector. So long as the private sector uses this money to save, it's creation is not inflationary and will not show up in the CPI. Inflation is caused by too many dollars chasing too few goods, and dollars under the mattress are not chasing anything.
There's an elegance to this perspective that comes with what is of course oversimplification. Winterspeak invites us to consider the government, the central bank, and the private banking system as a consolidated entity. Money taxed, lent to the government or deposited in a bank is money destroyed (at least temporarily). Money spent by the government, or borrowed from a bank is money created. Questions of control are not addressed — we don't know whether it is Congress, the Fed chair, or private bankers who most influence this public-private hybrid at the core of the monetary system. But thinking about money this way cuts through a lot of confusion.
Let's combine all this with the Moldbugian insight that money is the bubble that needn't break, that its value is due to the stability of a Nash equilibrium — as long as everyone wants it, it is rational for everyone to want it. The government/banking system, then, has the incredible power of creating what everyone wants or destroying it, as it sees fit via spending and taxation, but also via lending and borrowing. However, there may be constraints on its ability to use that power, for institutional and political reasons, but most profoundly because of the dynamics and potential fragility of Nash equilibria. We'll come back to this.
Both John Cochrane and Paul Krugman would agree that the current crisis, at least in part, is a phenomenon related to an unusual increase in people's desire to hold money. Moldbug's Nash equilibrium is on overdrive somehow: all everyone wants is what everyone wants, claims against the government that the government can create or destroy at will. To understand the implications of this phenomenon, we have to add a bit more substance to the meaning of "savings" and "investment". We'll follow Winterspeak's treatment first, which is remarkably insightful, although I don't entirely agree. (I hope W— will for give me for lifting such a large chunk of his post, but I'd not do it justice in paraphrase.) Winterspeak:
Essentially, Cochrane and Fama both assert that savings = investment (+ capital account), and so say that any Government stimulus will crowd out private investment. Here's the derivation (by identity) to get you S = I
Y = C + I + G
National savings can be thought of as the amount of remaining money that is not consumed, or spent by government. In a simple model of a closed economy, anything that is not spent is assumed to be invested:
NationalSavings = Y - C - G = I
If you think that banks make (investment) loans based on their deposits, then it's reasonable to assume that all money not spent (ie. saved) is invested. But banks do not take deposits and loan them out. In fact, banks make loans first and then those loans become deposits. Remember — loans create deposits, deposits do not "enable" loans.
Loans create deposits
Banks, by way of their Federal charter, can expand both sides of their balance sheet at will, subject to capital requirements. This money is created ex-nihilo, but always nets out to zero in the private sector, as each (private) asset that a bank creates must be matched by a (private) liability. Government can create money outside of the system, but banks always need to net out and balance the balance sheet.
People believe that fractional reserve banking, in some weird way, has banks taking deposits, multiplying it (through what seems like a strange and fraudulent process), and then making a larger quantity of loans. In fact, banks make whatever loans they think make sense from a credit perspective, and then borrow the money they need from the interbank market to meet their reserve requirements. If the banking sector as a whole is net short of deposits, it can borrow the extra money it needs from the Fed. If you think this is a weird and pointless regulation you are correct. Canada, for example, has no reserve requirements and yet seems to have a banking sector. The quantity banks can loan out is constrained by capital requirements and credit assessments.
Facts on the ground
If a description of how banks actually work doesn't shatter your belief that savings = investment, consider Reality. From about 2000-2006, American savings went negative, yet banks loaned out huge amounts of money (made huge investments). In fact, they came up with all kinds of clever ways to skirt capital requirements so they could make even more investments. If savings = investments, and savings fall, how can investments rise? By the same token, from 2006 to now, the private sector has actually delevered, saved, but banks aren't making any loans (investments). What's up with that?
More Facts on the ground
Anyone who thought they were saving by putting money in the S&P500 has had a rude wakeup call. They were not saving, they were investing, and now 40% of that money is gone. I don't think they will confuse saving with investing in the near future.
So, what is savings?
A better way to think of savings is to think of it as what's left after taxes, consumption, and investment.
Y = C + I + G
Net Private Savings = Y - C - I - T = G - T
Austrians will howl that it is unreasonable to define savings as a residual, there should be a term S for active savings, but people have to save in currency, and in a fiat, floating fx, non-convertible world, currency is not a store of value. Fiat currency trades bankruptcy risk for inflation risk, and fiat currency is all that's sitting in bank accounts. So the Austrians are right, there should be some way to actively save, but they are wrong, because fiat currency in a bank is not it.
You split out savings from investment and you get Net Private Savings = Government spending - Taxes, also known as the deficit. So, the Government runs a deficit (spends more than it taxes) in order for the private sector to have the extra money it needs, after consumption, investment, and paying those taxes, to net save. This idea totally blew my mind when I first encountered it, but it actually makes total sense.
So, if you acknowledge that savings does not equal investment, then you see that Government deficit enables private savings. This is the OPPOSITE of all the Chicago guys who argue that the Government deficit REDUCES national savings. Government is a currency issuer, why does it need to save? Does a bowling alley need to hoard the points it awards for strikes and spares? Everyone acknowledges that the Fed can print money, but few people actually think about what that means.
There is so much meat in this. First, a quibble. Winterspeak is disingenuous when he suggests that the disconnect between the miniscule savings rate and the maxicule investment rate argues against the traditional closed-economy identity S = I. The US very much was not a closed economy. The US was importing capital at a breathtaking rate to fund its investment boom, and globally the traditional S = I identity always held. In Bernanke-speak, a Chinese and Middle Eastern savings glut funded an American investment boom. That Americans saved less than they invested hardly mattered, as long as foreigners were willing to lend.
Winterspeak has not really disproved anything here. He has just derived a different definition of savings. The reason this is important is because we talk about savings, economists and real Americans alike slip between the two definitions unconsciously, which badly muddles things. On the one hand, we talk about the savings rate, which most certainly includes investment: If you were contributing to your 401-K or buying farm machinery, you were doing your part to keep the US savings rate above zero. But on the other hand, when we say that during this crisis, Americans' shift to savings is proving disruptive of aggregate demand, what we are talking about sudden desire to hold claims on money rather than to invest in real capital. Let's disentagle these two phenomena. Keeping with a closed economy, we'll define:
Investment = Y - G - C (Traditional savings)
ΔClaimsOnGovt = Y - C - I - T = G - T (Gov't deficit = Winterspeakian savings)
Actually, this is my party, and I hate how private expenditures are conventionally disaggregated into investment and consumption, while government expenditures are just "G". So, let PC be private consumption, PI be private investment, GC be government consumption, and GI be government investment. Then:
Investment = Y - GC - PC = GI + PI (Traditional savings)
ΔClaimsOnGovt = Y - PC - PI - T = GC + GI - T (Gov't deficit = Winterspeakian savings)
So here's a bit of insight: We can increase Winterspeakian savings in three ways: by increasing government consumption, increasing government investment, or reducing taxes. But note that an increase in Winterspeakian savings only results in traditional savings (current new investment) if the government purchases investment goods. I don't wish to take a side on the stimulus debate now, but I do want to point out that when people use the normatively charged word "savings" to imply investment, a tax reduction that enables purchases of bonds doesn't cut it. A tax reduction increases claims against the government without increasing aggregate investment. Individuals imagine they have saved, but collectively, we have only reorganized claims surrounding the consumption and investing we were already doing. (Arguably a tax reduction could saturate the demand for government claims and lead to more real private investment. But that's a dynamic scoring kind of story.)
Fundamentally, Winterspeakian savings is about restructuring our collective balance sheet in a way that leaves individuals with more claims on government. If the private sector reduces investment and the government increases investment to take up the slack, the ultimate result is the government controlling more real capital, and individuals holding claims on the government. In other words, what the private sector is doing right now is using financial markets to demand more socialism. Individuals no longer want to hold direct claims on private enterprise. They wish to hold claims on government, which implies the government must own and direct the productive activity that will enable it to make good on all those claims.
Regular readers may have noticed that I love telling ironic capitalists-are-socialists kind of stories. But I don't think a shift towards public ownership of the means of production is a good thing. A consumption-centered stimulus or tax cut would keep the government out of the business of investing while enabling people to hold more money, but that would lead to reduced future production despite an increase in "risk free" claims against government. Today's "savings" would be a losing investment, in real terms. That is, if we satisy the public's demand for an increase in government claims, but do not invest the proceeds well, we should expect a great inflation. Hopefully, we are capable of identifying sufficiently productive infrastructure and public goods investments that the government can earn a real return without interfering in the traditional private sphere. But in the intermediate term, there will be no substitute for increasing PI. We need to persuade individuals to undertake investments whose risk they hold and bear if we want to have a capitalist economy. We need a private financial system.
But did we ever have one? If you buy the Moldbug/Winterspeakian view of banking, we did not to the extent that individuals used the banking system to intermediate investment. Instead, the government effectively created incentives for bankers to invest in good projects by letting them keep the proceeds when they chose well, while punishing them a bit but then eating their losses if they chose poorly. We could reproduce that system more directly by having the government invest in private equity funds. Is that what we want?
We really need to think clearly about what we used to have, what was good and what was broken about it, and what we want going forward. How much should investment risk be socialized, to encourage entrepreneurship, vs how much should be borne privately, to encourage discrimination? Would it be possible to build the right mix transparently, rather than to rely on hidden guarantees and subsidies as we have until now? (It may be that subterfuge is prerequisite to an effective financial system in a political world. But I don't like to believe that.) To the degree that the investors will actually bear investment risk, can we define instruments that they can productively invest in? As index investors have learned, bearing risk without discriminating between good and bad is a prescription for disaster, eventually.
This has gotten terribly long, and terribly rambling, but I do want to come back to one idea. Early on, I described a Winterspeak/Moldbug synthesis in which the government comes off as incredibly powerful. Government money is what everybody wants because everybody wants it. The government destroys money through taxation and borrowing, and creates it by spending and lending, and can do so at will. A very large fraction of "private" lending is in the influence of government, by how it organizes the pseudoprivate banking system as well as via direct market interventions. With so much power, what are the government's constraints? Why can't it get an outcome that it wants, presumably a good economy with a bit of corruption on the side? Let's put aside the institutional stuff. (For example, governments usually have to pretend private banks are private; their ability to direct the loans they guarantee is limited; in fact, a bankers-control-government-expenditure story is historically more compelling than government-controls-bank-expenditure). Right now, the government can pretty much do what it wants with the banking system, can expand the quantity of risk-free claims to meet demand at will, can lend to whomever it pleases. Why can't it fix the country?
Cassandra had a brilliant post not long ago called "an idea crunch". Read it if you haven't. Ultimately, a financial system has to find productive projects for the private parties to invest in. The government can invest directly, can delegate investment to the best and the brightest, can saturate the public's demand for money until private parties try to find other means of storing wealth. But it's what real human beings do with real resources that ultimately matters. Our financial system didn't fail because it was overlevered. It failed because it was uncreative: It could not conjure up worthwhile things to do with the capital it was asked to invest, and instead of owning up to that, it pretended that poor projects were good. Financial markets are ultimately information systems. The only way out of this is to discover worthwhile things to do, or more importantly, to develop better means of generating a diverse menu of worthwhile things to do going forward. Right now, the government is being asked to do what the semi-private financial system could not: generate a positive real return on trillions of dollars of undifferentiated future claims. The stakes are very high — that Moldbugian monetary Nash equilibrium can be a bitch. Money is the bubble that need not pop, but that's no guarantee that it won't. Anything that's desirable only because everybody desires it is just a single major failure away from being yesterday's darling. If unthinkable banking system failures can occur, so can unthinkable failures of the monetary system.
 I think Cochrane's essay has been too roughly treated in the blogosphere, and I don't wish to pile on. A fair reading of the piece makes it clear that Cochrane is not making the elementary errors that an excerpt from the beginning might suggest. I quote Cochrane's essay here with some affection. I think his view of money is wrong, but like nearly all of economics it is better read as a form of organized hope that the world might be rendered coherent than as a reliable description of the world as it is.