I am addressing the general issue of why two lists might have different values for the GI for potatoes, and ignoring any mechanism for how boiling or baking may affect the GI.
The story is far more complicated that these simple lists might suggest.
The South Beach Diet Plan's Web-Site list provides no references, so that is the first strike against it.
Meanwhile, the Harvard Health Publications list references International Tables of Glycemic Index and Glycemic Load Values: 2008 from
Diabetes Care December 2008 vol. 31 no. 12 2281-2283, doi: 10.2337/dc08-1239.
In that article, the complexities start becoming clearer.
Some research laboratories continue to use white bread as the
reference food for measuring GI values, whereas others use glucose (dextrose); therefore, 2 GI values are given for each food.
The first value is the GI with glucose as the reference food (GI
value for glucose = 100; GI value for white bread = 70), and the
second value is the GI for the same food with white bread as the
reference food (GI value for white bread = 100; GI value for glucose = 143). When bread was the reference food used in the original study, the GI value for the food was multiplied by 0.7 to
obtain the GI value with glucose as the reference food.
So, if we assumed that the Harvard list was correct at 82, and used glucose as the reference, but the South Beach Diet used bread as the reference, we would expect it to give a value of 57 - that is pretty consistent with the 56 it gave. There is no need to assume either is wrong just that one is using a different (older) definition of the scale than the other.
However, even having one single value is confusing. Sticking to the glucose scale, this is just some of what the Diabetes care journal includes for potatoes:
- Ontario, white, baked in skin (Canada) 60
- Russet, baked without fat 56
- Russet, baked without fat, 45–60 min 78
- Russet, baked without fat 94
- Russet, baked without fat 111
- Desiree, peeled, boiled 35 min 101 ± 15
- Nardine 70 ± 17
- Ontario, white, peeled, cut into cubes, 58
- Pontiac, peeled, boiled whole for 30 min 56
- Pontiac, peeled, boiled 35 min 88 ± 9
- Prince Edward Island, peeled, cubed, boiled in salted water 15 min, 63
- Sebago, peeled, boiled 35 min 87 ± 7
- Boiled or cooked, white or type NS 24
This list continues much further, I just got bored of transcribing. You might like to take a look at it to look up the closest type of potatoes to the ones you consume, and how you prepare them. They also have far more detail (including some of the missing error ranges, when the original test was done with the bread scale), references to the original studies, sample sizes, etc. Where several studies are comparable, they also provide grouped averages.
In conclusion, it is unreasonable to expect that a wide-range of naturally grown potatoes, from many different countries, and measured by the effect on human volunteers, to have a single, precise value for their GI index.