To his credit, Paul Ryan did not once invoke "bootstraps." In a speech unveiling his new poverty proposal Thursday morning at the conservative American Enterprise Institute, he did not describe the safety net as a "hammock" a single time. He did not use thinly veiled code words about the "inner city." He did not misleadingly suggest — as he has often in the past — that the war on poverty has failed, when in fact much research tells us that federal programs, despite their flaws, have lifted millions of Americans out of destitution.

Ryan is learning to talk about poverty in ways that don't make his ideas instantly unpalatable to the left, to advocates for the poor, to researchers and program officers who recognize that poverty is a sign of so much more than the absence of effort. But for all of his welcome nuance — his new plan includes several ideas that can be cheered across the political spectrum — Ryan still misunderstands or glosses over some crucial points about the causes of poverty and what's needed to alleviate it.

The centerpiece of Ryan's deficit-neutral proposal is an idea to consolidate the federal government's many anti-poverty programs, including food stamps, cash assistance and housing vouchers, into a single more flexible funding stream that would be available to the states in the form of block grants. Local public or private service providers who know best would then tailor those resources to the needs of individual recipients. Maybe one family needs housing help. Maybe another requires subsidies for child care. This idea rightly recognizes the poverty isn't experienced the same way by everyone. But here is what Ryan's plan suggests should happen next:

Providers must be held accountable, and so should recipients. Each beneficiary will sign a contract with consequences for failing to meet the agreed-upon benchmarks. At the same time, there should also be incentives for people to go to work. Under each life plan, if the individual meets the benchmarks ahead of schedule, then he or she could be rewarded.

His "discussion draft" says more about the benefits to achieving these goals (like "getting a job within six months") than the sanctions for failing to meet them. But the idea is fundamentally punitive. It betrays the fact that Ryan's latest thinking has not strayed all that far from the simplistic notion that people in poverty are solely to blame for their own circumstances. An incentive system like this assumes that end goals such as employment are entirely within the control of poor people if they would just try hard enough.

This notion fails to recognize that, while personal responsibility is important, so too are structural obstacles in the economy, in the education system, in the housing market. We can hardly expect personal effort alone to overcome poverty without systemic investment on society's part on the fronts beyond a poor person's control.

The idea of a contract with punitive benchmarks also ignores lessons that researchers have learned about the effects of poverty on cognition. Princeton psychologist Eldar Shfair and Harvard economist Sendhil Mullainathan have argued that the stress of living in poverty sucks up mental bandwidth the rest of us take for granted. That mental tax means that a mom may forget to take her medication, or to pay a utility bill on time, with each mistake yielding a cascade of other problems. This means that living in poverty is like living without room for error. It also means that we should design anti-poverty programs that are flexible and forgiving, not punitive and deadline-oriented.

Another crucial element of Ryan's plan is that we should test results as states try out new programs, replicate what works and ditch what doesn't. In theory, evidenced-based policy is an important idea, and it's one everyone can agree on. But, particularly in the realm of poverty, measuring success is much more complicated than Ryan suggests (his proposal devotes just one page to this topic).

Poverty is inherited across generations, which means that success overcoming it is a long game, too. Measuring it is not just about having access to data; it's about having patience in collecting it. It's about finding rigorous ways to isolate the value of interventions among the many interconnected factors that contribute to and stem from poverty. How are we to know, for instance, if a child performs better in school because she was given extra tutoring, or because she was given nutrition assistance, or housing stability?

How will we know what outcomes we're looking for? The literature is full of unexpected connections: food stamps improve graduation rates, health care access increases college enrollment, housing stability leads to children with better employment prospects as adults. If we're going to look for "results," then we must acknowledge that aid to the poor can yield positive outcomes that are only visible when we look beyond family income and employment.

Here is one final complication: The appeal of Ryan's plan is that it would grant local officials greater control to tailor solutions to the kinds of problems their poor residents face. Perhaps housing affordability is a bigger issue in San Francisco. Maybe reliable transportation is an acute problem in Atlanta, or childcare access an obstacle in New York. Allowing local governments greater freedom here recognizes that different communities, just like different families, have varying needs.

But here is another way to look at this variation:

That's a map of states ranked by a cumulative index of child well-being, from an Annie E. Casey Foundation report I wrote about Wednesday. This is a Census map of poverty rates in the United States:

This is a map from the Kaiser Family Foundation of states that have refused the Medicaid expansion in the Affordable Care Act:

This is a Census map of the share of adults by state, over 25, with more than a high school degree:

In reality, some states have worse track records — and differing commitments — to caring for the poor, to providing them health care, to lifting children and families out of poverty, to educating them. While block grants would recognize that a state like Texas has different needs from Minnesota, it would also place greater control in combating poverty in the hands of local governments with weak records on this front.

A child's chances of moving up in the U.S. are already partly determined (and constrained) by where he or she lives. How would that picture change if we turned over much of the challenge of confronting poverty to the states themselves, with their differing views on who's "deserving," which needs are "basic," and what role government should play? Ryan says he wants to start a conversation about reforming poverty programs. These questions should be a part of it.