Zoron :
Notice I said Intel's 7th generation Pentium processor... not their 7th generation processor period. I am fully aware that Nehalem is really their 10th generation processor...
I don't think it is their 7th-generation Pentium processor, either.
By architecture:
1st-generation Pentium = P5 Pentium
2nd-generation Pentium = Pentium Pro
3nd-generation Pentium = Pentium II
4rd-generation Pentium = Pentium III
5th-generation Pentium = Pentium 4 i686, 20-21 stages (Willamette/Northwood)
6th-generation Pentium = P6+ Pentium M/Pentium Dual Core (Yonah)
7th-generation Pentium = Pentium 4/D EM64T, 34 stages (Prescott/Smithfield, Cedar Mill/Presler)
8th-generation Pentium = Core-based Pentium Dual Core (Merom-2M/Allendale)
That would put Nehalem as a ninth-generation Pentium if any Nehalem-based chips were sold under the Pentium moniker.
By brand name:
1st-generation Pentium = P5 Pentium
2nd-generation Pentium = Pentium Pro
3nd-generation Pentium = Pentium II
4rd-generation Pentium = Pentium III
5th-generation Pentium = Pentium 4
6th-generation Pentium = Pentium M
7th-generation Pentium = Pentium D
8th-generation Pentium = Pentium Dual Core
That would also put Nehalem as an ninth-generation Pentium.
About the only way I can see Intel calling the Nehalem a seventh-generation Pentium is if they ignore the Pentium Pro as it is a server processor and lump together the pretty-dissimilar P4/Pentium D revisions as all being NetBurst and one generation. They could also lump together the PPro, PII, and PIII all under the heading of "P6," in which they should have named the PII "Pentium Pro w/MMX" and the PIII "Pentium Pro w/SSE" as they did with the original Pentium vs. Pentium MMX processors. That would have made 32-bit NetBurst the third Pentium revision, P6+ the fourth revision, EM64T NetBurst the fifth revision, and Core the sixth revision.
Nothing necessarily wrong with Core 3... but then I'm not the one making the decisions at Intel. I only said the name made sense, not that it was good.
I agree their naming schemes have been poor ever since the Pentium M in 2003. Most people caught on that the Pentium M had nothing to do with the original P5 Pentium, but this was the start of the ambiguous naming schemes. The Pentium M was for "Mobile" as was the Celeron M. That's fine. But when the D in Celeron D was for "Desktop" and the D in Pentium D was for "Dual-core," people got quite confused and for good reason. It got much worse when the Core 2 came out. The Core Solo and Core Duo were easy enough for people to understand as the Core Solo was a single-core chip and the Core Duo was a dual-core chip. The Core 2 confused the snot out of people at first as quite a few couldn't tell how many cores or CPUs they had based on the name. I remember hearing the "does the Core 2 mean it has two cores?" and "Does a Core 2 Duo have two dual-core processors in it?" bit a lot in late 2006 until people understood that Core 2 was the brand and had nothing to do with the number of cores in the chip. And then try to explain to a new techie that Core chips use the P6+ uArch while Core 2 chips use the Core uArch. I'll betcha a bunch of people will think that a "Core i7" has seven cores at first and be bewildered at finding out it actually has two or four cores.
Personally, I'd have used a new name that was unambiguous. Pentium 5 would have been a natural as people know that Pentium and then a number is the brand. The "i-and-a-number scheme would have been great if the number would have been the number of cores. And then do what they used to do and prefix it with "mobile" if it is a mobile chip:
Pentium 5 i2 = Nehalem-based dual-core desktop chip
Pentium 5 i4 = Nehalem-based quad-core desktop chip
Pentium 5 i8 = Nehalem-based eight-core desktop chip
Mobile Pentium 5 i2 = Nehalem-based dual-core laptop chip
Mobile Pentium 5 i4 = Nehalem-based quad-core laptop chip
This extends well for newer chips as well:
Pentium 6 i2 = Sandy Bridge-based dual-core desktop chip
etc.