Just before the election last fall, Elon Musk predicted that he could cut $2 trillion from the federal budget. By early January, a couple of weeks before President Donald Trump returned to the White House and appointed Musk head of the newly created not-quite-department dubbed the Department of Government Efficiency, he was calling $2 trillion "the best-case outcome" and saying "we’ve got a good shot" at getting $1 trillion in cuts. In late March, he said DOGE would deliver $1 trillion in deficit reduction by the end of May. By April 10, that was down to savings of $150 billion. Draw a trendline through Musk’s pronouncements so far, and it shows the DOGE savings number going negative on May 21 and becoming a deficit increase of more than $2 trillion on Dec. 1.

This is not meant as a serious forecast. But with Musk’s time as a "special government employee" due to end this month, and the Partnership for Public Service estimating that the cost in lost productivity, rehiring of fired workers and other side effects of his chaotic approach will top $135 billion this year, it does seem fair to conclude that the direct annual savings from DOGE’s efforts will be less than $150 billion, which amounts to about 2% of federal spending. And with spending up 10% since President Trump took office relative to the same period last year; the "one big beautiful bill" addressing taxes and spending currently working its way through Congress almost certain to increase the deficit; and a tariff-induced, tax-revenue-reducing recession perhaps on the horizon, I guess it’s not impossible that deficit will rise by $2 trillion.

Its minimal impact on spending notwithstanding, DOGE has had a huge impact in Washington, with its actions disrupting the federal government in ways that few would have imagined possible before it set to work. What, then, have Musk and DOGE really been trying to do?

There are many, many theories on this. In March, Alex Nowrasteh and Ryan Bourne of the libertarian Cato Institute tried to organize them into "six theoretical models for understanding DOGE’s action to date":

Nowrasteh and Bourne offered their own evaluations of how well these models were supported by the facts on the ground — I highly recommend their essay. But six weeks have passed since they wrote it, so here’s my updated run-through, plus a suggestion for an addition.

DOGE and others in the Trump administration have made a point of targeting agencies (USAID, the Department of Education) and functions (anything with diversity, equity or climate in the title) that are seen as staffed by left-leaning bureaucrats or providing funding to left-leaning scholars and nonprofits. I have some sympathy for the view that elected officials shouldn’t have to rely on career bureaucrats who oppose their priorities. The ways DOGE has gone about getting rid of people and programs, though, have in many cases clearly violated the law. There’s a chance the Supreme Court will strike down some of those laws as unconstitutional, but if it doesn’t, President Trump may be left with a federal workforce — which before he entered politics was pretty evenly split between Democrats and Republicans, but seems to have become more Democratic since — that’s positively seething at him.

At Tesla Inc., SpaceX and X.com, Musk has been in the habit of shutting down key functions or features and then reviving them if there’s enough outcry — a messy but at times quite effective way of figuring out what truly matters. The federal government is not a corporation, and I don’t think its feedback mechanisms work the same way. The jury’s still out on this one — maybe, as my colleagues at Bloomberg’s editorial board write, a saner attempt at civil service reform will arise from DOGE’s wreckage — but I’m not super optimistic.

Theories 3 and 5 contradict each other, so let’s consider them together. Most federal spending, and almost all long-run spending growth, is in so-called entitlement programs such as Social Security, Medicare and Medicaid, not in the discretionary programs for which Congress has to appropriate money each year and that have been the main targets of DOGE’s cuts. White House Office of Management and Budget Director Russell Vought has argued that discretionary cuts are still important because Americans "are simply not going to buy the notion that their earned entitlements must be tweaked while the federal government is funding Bob Dylan statues in Mozambique or gay pride parades in Prague." That may be right, but Vought’s hopes for entitlement reform face a possibly insurmountable obstacle in the form of his boss. President Trump has said repeatedly that he opposes cuts to Social Security, Medicare or Medicaid, hinting in a recent Time Magazine interview that they were unnecessary because his administration was rooting out so much "waste, fraud and abuse." The safest bet, then, is on more fiscal irresponsibility.

Vought has been driving this, promoting the legal argument that when Congress appropriates money for something, the president can’t spend any more than that amount but isn’t obliged to spend anything at all. Legal scholars seem to think the Supreme Court won’t buy it, but, hey, maybe it’s worth a try.

The staff of the Democratic minority on the Senate Permanent Subcommittee on Investigations has reported that as of Jan. 20, the day Trump was sworn in, "Musk and his companies were subject to at least 65 actual or potential actions by 11 different federal agencies" and that the potential liability from the 40 of those actions for which it was able to make estimates added up to $2.37 billion. A new biography speculates that Musk’s anger over National Transportation Safety Board investigations into Tesla helped spur his reinvention as a right-wing activist and then president-whisperer. So, yes, it’s plausible that Musk has been mainly just looking out for his own interests in Washington, something that federal conflict-of-interest laws are supposed to prevent. But it also doesn’t seem to be working out very well for him, with Tesla sales plummeting because of backlash over his political activities and President Trump’s tariff increases threatening his businesses even further.

The most obvious gap to me in Nowrasteh and Bourne’s lists is the DOGE priority that, especially as Musk’s involvement winds down, could be most consequential. It’s most easily described in negative terms — "DOGE is building a surveillance state" was the headline on a recent New York Times piece by veteran tech journalist Julia Angwin. When I ran this by Nowrasteh, he suggested that it might best be understood as a subsidiary of theoretical model No. 1, in the sense that "DOGE is a way to enforce conservative policy priorities such as using the full power and data owned by the federal government to identify and deport illegal immigrants." It has definitely been trying to do that, uploading sensitive data from multiple agencies to create a "master database" for immigration enforcement. But I think the project is broader than that. To borrow from the executive order that created it, DOGE is "modernizing Federal technology and software to maximize governmental efficiency and productivity," in particular by "promot[ing] inter-operability between agency networks and systems."

Information has had a tendency to stay siloed within different federal agencies. That’s partly because the computer systems are old and incompatible and bureaucrats are set in their ways, but also because it aids in keeping data secure and protecting Americans’ privacy. Given the mix of malice and incompetence that has already become something of a trademark of the second Trump administration, it’s hard to see DOGE’s efforts to break down these barriers in anything but a negative light. But I think back to Airbnb co-founder and DOGE volunteer Joe Gebbia’s hopeful words in February about wanting to "improve the user experience within our government," and wonder if maybe, just maybe, with the hyperpolitical Musk mostly out of the picture, DOGE could actually bring some positive changes. Then again, there’s been a steady undercurrent of speculation that the real endgame of DOGE is just to replace humans in government with artificial intelligence — maybe that should be theoretical model No. 7 — which is less reassuring.

This column reflects the personal views of the author and does not necessarily reflect the opinion of the editorial board or Bloomberg LP and its owners. Justin Fox is a Bloomberg Opinion columnist covering business, economics and other topics involving charts. A former editorial director of the Harvard Business Review, he is author of "The Myth of the Rational Market."