In 1989, the Supreme Court ruled that the jury instructions that Texas was using were constitutionally deficient because they failed to ensure that jurors could give meaningful consideration to a defendant’s mitigating evidence. Under the Texas system, jurors were instructed to respond to only two questions: whether the killing was deliberate, and whether the defendant posed a continuing threat to society. If the answers to both were yes, a death sentence was automatic.

The Texas Legislature addressed the problem two years later by instructing jurors to take “all of the evidence” into consideration, including the defendant’s character and background. But in the interim, during which Mr. Smith was sentenced to death for murdering a former co-worker at a Taco Bell in Dallas, judges tried to address the problem by telling jurors that if they thought the mitigating evidence warranted a sentence of life in prison rather than death, they should simply answer no to one of the two questions, even if they believed that the proper answer was yes.

The Supreme Court again called a foul and sent the case back to Texas:

Eventually, the Supreme Court held that this “nullification instruction” was constitutionally inadequate as well. It applied that ruling to Mr. Smith’s earlier appeal, overturning his sentence in 2004 and sending it back to the Texas Court of Criminal Appeals, which in turn reinstated it, finding the error “harmless” because Mr. Smith had failed to show that the nullification instruction had caused him “grievous harm.”

The defense argument yesterday:

In his new appeal, Mr. Smith, represented by a University of Texas Law School professor, Jordan M. Steiker, is arguing that the state court’s resolution of the case flew in the face of the Supreme Court’s analysis. The state court, having failed in the first round to apply its “harmless error” rule, should not be permitted to introduce it after the fact, Mr. Steiker said.

ScotusBlog has more, as does Capital Defense Weekly.