We hear lots of talk in the business community about the need to innovate to stay ahead of the competition. But what does “innovation” really mean? Merriam-Webster defines it as “a new idea, device, or method.” If Trump and Kushner intend to import the private sector’s ideas, devices, or methods into government, we should take a closer look at how innovation works in the business world.

The federal government is a large organization; in fact, it is the largest employer in the U.S. followed by Walmart. So if innovation is going to work in the federal government, it would need to follow the same model of innovation at large private sector organizations.

But the tough truth is that it’s really hard to innovate in large private sector organizations. There are really only two ways to do so effectively: acquire those that innovate or innovate internally.

Large private sector companies have optimized their structures and operations to execute on known business models. They are good at cutting costs and creating efficiencies when they have figured out these business models. Startups, on the other hand, are often trying to find their business models. They have to be agile and experimental before they run out of money. Once they do, they are often acquired by large companies who then turn the crank and optimize them to scale.

What about the other way to innovate—internally? That’s incredibly difficult. It has to be done through experimental innovation projects, which have to be protected from the rest of the organization and measured and evaluated differently than other business functions. The culture of many large organizations is not welcoming to innovation, since most have optimized their operations and reward systems around doing predictable work over and over again.

Therefore, trying to apply private sector innovation to the public sector, with its overlapping and sometimes conflicting legislative mandates and a bureaucracy organized to execute on those mandates, is a gargantuan task. If Trump and Kushner want Washington to innovate internally, they’ll have to begin with experimental and small-scale efforts that large private companies have tried in the past.

For example, Lockheed Martin set up its Skunk Works program with a license to break rules and innovate, and it resulted in several successful aircraft designs. Google’s policy of giving engineers 20% of their time to work on a company project that interests them led to the development of Gmail and Google News. But providing employees with space to think takes time and doesn’t have the immediately visible payoffs often required with public-sector projects.

As for the other option, the government can’t “acquire” innovative startups like private companies can. So in order to have any immediate impact, Washington would need to outsource specific operations to the private sector. These would need to be well-defined tasks that can be tangibly measured, such as IT operations.

Kushner would also have to figure out how to downsize the government bureaucracy that is being outsourced. He won’t be able to easily spin off those people to the private sector companies that take over the outsourced functions. Furthermore, he’ll likely face legal and technical difficulties and a large publish backlash if he tries to massively lay off government workers.

An additional complication is that the new government contracts are likely to go to large private organizations that already know how to navigate the government contracting process. Given the bureaucratic realities of the current government contracting system, which is subject to public sector legal mandates and union rules, we’re not likely to see the same truly innovative approaches that unencumbered startups employ.

Kushner’s best option is to figure out how to entice the entrepreneurial world to collaborate with government. Some large private companies use their size as a platform to attract and showcase innovative approaches. For example, Walmart is acting as a startup incubator of sorts, convening and supporting small companies to develop technology for retail as it transitions from brick-and-mortar sales. This process is exposing the Walmart bureaucracy to new possibilities and creates a positive environment for trying out new ideas—and in the process beginning to change the company’s mindset. Perhaps the same can be done with the federal government.

Joe Hadzima is president and co-founder of IPVision and a senior lecturer at the Trust Center for MIT Entrepreneurship at the MIT Sloan School of Management.