The author assumes 1.5% leakage rate. Natural gas itself emits less CO2 when burning however methane is substantial contributor to GHG. The author concludes there is very little data on amount of natural gas leakage.
If we assume a 1.5% leakage rate, this would have a greenhouse gas warming potential equal to
14.8 g C of CO2 per million joules of energy. This would be additive to the emissions during combustion
(13.7 g C of CO2 per million joules of energy) and to the emissions associated with obtaining and
transporting the natural gas (very roughly estimated above as 4.5 g C of CO2 per million joules of
energy). Total greenhouse gas emissions from natural gas from hydraulic fracturing may, therefore, be
equivalent to 33 g C of CO2 per million joules of energy. For diesel fuel or gasoline, the total greenhouse
gas emissions are equivalent to approximately 20.3 g C of CO2 per million joules of energy.
If leakage was say 1% then it would be like 28g per MJ and 2% would be even worse at 38g per MJ.
Also the author doesn't account for long distance transportation of oil. The lifecycle CO2 from gasoline will depend on length of transportation. As we use up more and more "local" oil we will need use more and more "long distance" oil and CO2 cost will rise.
Last thing to consider is efficiency of powerplant. ICE are about 15% efficient (tank to wheel). Using a fuel cell instead of internal combustion engine for natural gas could cut CO2 release per mile by half or more.
Internal Combustion Gasoline: 20.3g / 0.15 = 135g per MJ effective work (work at wheels).
Internal Combustion Natural Gas: 33g / 0.15 = 220g per MJ effective work (work at wheels).
Fuel Cell Natural Gas: 33g / 0.40 = 82.5g per MJ effective work (work at wheels).