Yes, these battery life measurements are quite impressive. As noted earlier, the battery capacity is ~ 30% higher on the LG G2 vs. the HTC One. Normalizing for battery capacity, the G2 has 16.8% better 3G web browsing battery life, 5.5% better WiFi web browsing battery life, and 28.1% better cellular talk time battery life than the HTC One. So in addition to TSMC's 28nm HPM fabrication process, LG is using something called Graphics RAM [GRAM] that, when the display is static, "the G2 can run parts of the display subsystem and AP off and save power".
I can't help but think that reviewers are not able to capture true power efficiency when performing their battery life tests. To accurately measure power efficiency, one needs to measure both power consumption and performance. Measuring power consumption without measuring performance (and vice versa) doesn't give us the full story. Given the difficulty in easily measuring application-specific power consumption on handheld devices, there may be no good solution to this at this point in time.
On a side note, the G2's GPU performance is very good for a phone but is anywhere between ~15-25% lower than Qualcomm's MSM8974 MDP tablet reference platform, which just goes to show how these high performance SoC's are not able to stretch their legs in this form factor.
