This VGChartz article bugged me as an example of bad and sometimes contradictory advice.

Lesson 1 - Don't Take Six Years to Make Your Sequel
Gran Turismo 4: 89.61%
Gran Turismo 5: 84.69%

First of all, a MetaCritic GameRankings score (why is he using GameRankings instead of Metacritic, anyway?) drop of 5% is not necessarily statistically significant, especially since there's no way, from just the scores, of verifying that the same selection of outlets or reviewers were included. The drop may indicate a drop in popularity of the genre, or many factors other than the development cycle.

If what he means to say is that the earlier you release, the earlier you get paid, then he's right, but for every GT5 that fails to meet expectations or Daiktana that becomes the butt of jokes or DNF that fails to materialize, there are games rushed out the door lacking polish or originality. Of course, since the subject is sequels perhaps originality is not a factor.

Sure, six years might be extreme, and there might very well be a point of diminishing returns, where more development time doesn't necessarily improve things. However there may very well be cases where 2 years is not enough, and where even three or four may be justified. Might not another 6-12 months have improved, say, Halo 2?

Lesson 2 - Don't Re-Use the Same Engine Over and Over Again
Fallout 3 (X360): 92.79%
Fallout: New Vegas (X360): 83.64%

This potentially conflicts with #1, given that the more changes you make to an engine, the more the engine development cycle impacts your content development cycle, and the longer the cycle gets. That's why DNF isn't out yet.

The choice of Fallout 3 and Fallout 3: New Vegas is also disingenuous to illustrate the idea that you need to iterate your engine. Those are the last two games to use Gamebryo, and one is essentially a full-price expansion pack for a fairly recent game. The Metacritic score for F3NV may be a better indicate for the objective age of Gamebryo, rather than just the idea that you shouldn't re-use engines, or the fact that it shares a lot in common with Fallout 3 itself, such that some people didn't see it as all that impressive.

Here's a more valid comparison:

Oblivion: The Elder Scrolls 4 (X360): 93.79%
Fallout 3 (X360): 92.79%

Oblivion and Fallout 3 are both Gamebryo engine games, although as usual, some improvements were made, so the later game does look a bit better, just like Oblivion looks better than Morrowind (at least from a purely technical perspective).

They are also based on two completely different premises and fictional universes, and were released 2 years apart. Still plenty of time to lament that the game didn't use the latest whizbang engine-- and only a 1% drop in score. The GOTY edition that came out a year later scores 95% on Gamerankings, coming out a year later than the initial release and a percentage point higher than Oblivion.

Sure, don't keep using the same version of the same engine for five or ten years (although frankly there are plenty of old school games that I'd buy new content for even a decade after initial release or more). On the other hand, though, don't spend all your development time and dollars on pushing pixels and polygons in pursuit of the best available buzzword compliance, either. Sometimes Good Enough is good enough when it comes to technology.

Halo 3: ODST (like Fallout 3: New Vegas) used essentially the same engine. I don't know to what degree Bungie enhanced that engine for ODST, or to what extend Obsidian did not modify the Fallout 3 code, but the essential concept is the same: a little bit more of the same using the foundation of existing technology. Is a bit more of a good thing, for the people who want more, such a bad thing?

Oh, and while we're on this point, don't let another developer use your engine. Ever.

I'll admit that not every developer is up for the task of documenting, licensing, and supporting the code they make to do their own games. Id used to do quite a lot of it, and certainly Epic does now (lawsuits notwithstanding). However, advising developers to not license their engine (and conversely telling everybody to reinvent the wheel since if nobody is licensing engines, you can't license one) is also bad advice. Looking at the buggy mess that is F3NV is cherrypicking.