531US1 126 Unit: $$U9 [05-01-02 10:02:37] PAGES PGT: OPIN BUSH v. GORE Stevens, J., dissenting Admittedly, the use of differing substandards for determining voter intent in different counties employing similar voting systems may raise serious concerns. Those concerns are alleviated—if not eliminated—by the fact that a single impartial magistrate will ultimately adjudicate all objections arising from the recount process. Of course, as a general matter, “[t]he interpretation of constitutional principles must not be too literal. We must remember that the machinery of government would not work if it were not allowed a little play in its joints.” Bain Peanut Co. of Tex. v. Pinson, 282 U. S. 499, 501 (1931) (Holmes, J.). If it were otherwise, Florida’s decision to leave to each county the determination of what balloting system to employ—despite enormous differences in accuracy 4—might run afoul of equal protection. So, too, might the similar decisions of the vast majority of state legislatures to delegate to local authorities certain decisions with respect to voting systems and ballot design. Even assuming that aspects of the remedial scheme might ultimately be found to violate the Equal Protection Clause, I could not subscribe to the majority’s disposition of the case. As the majority explicitly holds, once a state legislature determines to select electors through a popular vote, the right to have one’s vote counted is of constitutional stature. As the majority further acknowledges, Florida law holds that all ballots that reveal the intent of the voter constitute valid votes. Recognizing these principles, the majority nonetheless orders the termination of the contest proceeding before all such votes have been tabulated. Under their own rea4 The percentage of nonvotes in this election in counties using a punchcard system was 3.92%; in contrast, the rate of error under the more modern optical-scan systems was only 1.43%. Siegel v. LePore, 234 F. 3d 1163, 1202, 1213 (charts C and F) (CA11 2000). Put in other terms, for every 10,000 votes cast, punchcard systems result in 250 more nonvotes than optical-scan systems. A total of 3,718,305 votes were cast under punchcard systems, and 2,353,811 votes were cast under optical-scan systems. Ibid.