Only mildly confounded, and the difference is only truly pronounced at larger orders. Given a complicated enough conditional in a block though and this could have a serious effect.

factor_conditionals.rb

[24] pry(main)> def test(range, gt=nil)                                                                                                                                                                        
[24] pry(main)*   range.select { |n| gt ? n > gt : n }.reduce(:+)                                                                                                                                              
[24] pry(main)* end  
=> nil
[25] pry(main)> def test2(range, gt=nil)                                                                                                                                                                       
[25] pry(main)*   filter = gt ? -> n { n > gt } : -> n { n }                                                                                                                                                   
[25] pry(main)*   range.select(&filter).reduce(:+)                                                                                                                                                             
[25] pry(main)* end  
=> nil
[26] pry(main)> Benchmark.measure { test((1..1_000_000)) }.real                                                                                                                                                
=> 0.208830022
[27] pry(main)> Benchmark.measure { test((1..1_000_000),100) }.real
=> 0.213170894
[28] pry(main)> Benchmark.measure { test2((1..1_000_000)) }.real
=> 0.201729584
[29] pry(main)> Benchmark.measure { test2((1..1_000_000), 100) }.real
=> 0.198336247
[31] pry(main)> Benchmark.measure { 100.times { test((1..1_000_000)) } }.real
=> 18.096055603
[32] pry(main)> Benchmark.measure { 100.times { test2((1..1_000_000)) } }.real
=> 15.557792238

What Ruby version are you using to test this?

These are my results:

[4] pry(main)> Benchmark.measure { test((1..1_000_000)) }.real
=> 0.186278
[5] pry(main)> Benchmark.measure { test((1..1_000_000),100) }.real
=> 0.209084
[6] pry(main)> Benchmark.measure { test2((1..1_000_000)) }.real
=> 0.175825
[7] pry(main)> Benchmark.measure { test2((1..1_000_000), 100) }.real
=> 0.179605
[8] pry(main)> Benchmark.measure { 100.times { test((1..1_000_000)) } }.real
=> 18.893062
[9] pry(main)> Benchmark.measure { 100.times { test2((1..1_000_000)) } }.real
=> 17.401746

There's still a difference, but it's less pronounced.