I fear there is no simple information we can give here that explains where exactly it gets hard.

The situation differs depending on many different things.

Is porting a game as simple as just changing the code from a particular programming language to another language?

There is nothing simple about moving from one language to another. More often than not you need completely different approaches with completely new libraries etc. I think in most cases the language itself stays the same, but even then porting can be extremely hard if keeping the codebase portable was not one of the original goals. Unless you have many abstraction layers and people who can quickly write new implementation for all the features you pretty much start from scratch.

Even if the language stays the same, people might have specialized on some APIs (application programming interface) that exist for a console.

If they have to program on a very low level things like the architecture / instruction set of the CPU might make a difference as well.

You can read articles about software portability to get an idea of the challenges that exist in that area.

The only factor that I read in regards to the question I have was from a developer stated in an interview as having lack of experience developing pc games.

It is also important to know that for consoles the specs are set in stone while you need to support several different hardware constellations when you develop for desktop pcs.

That might be the main reason behind the statement - a lack of experience in that area.

Games are created on computers for sure, but its not always the case that there's a PC version running behind the scenes. In general, I'd wager that many studios who don't start out to publish a PC version, simply don't have one at all; of the ones that do maintain one, its either because their engine/middleware make it trivial, or they take effort to maintain a sort of "minimal-effort" port who's primary purpose is for debugging -- they certainly aren't optimizing their PC build if they never plan to ship it.

As to why its hard to port from one to the other after the fact, its because PCs and consoles work in fundamentally different ways; On the PC, you often have greater hardware resources, but you also pay a much larger cost of abstraction. On the Console, you have less hardware to throw at the problem, but you can take tight control of it without having to negotiate 3-4 layers of abstraction. We're also at a point now where the two higher-end consoles both have 8GB of RAM, and where 8GB is not quite the minimum that all PCs are sold with even today, let alone the perfectly-good computer you bought 12-18 months ago -- and even that fails to account for sharing it with other programs, which the consoles don't really do either.

On the PC platform, you have to make the general case go fast on somewhat "better" hardware (typically, faster CPU, faster GPU, but same or less RAM), while on the console you have to make a couple very specific cases go fast on somewhat odd, if not under-powered, hardware (more but slower CPUs, a middling GPU, a reasonable amount of RAM.

PC ports can suck because the PC is a very different beast than a console.

There's always a learning curve, sure, and Stainless laid out the general flow of title quality pretty well. But I think the launch-window-titles we see this generation will prove to be better representatives of the consoles' potential than launch-windows titles of generation's past. The PS2 launch titles were notoriously poor, for example. Already this generation we see some fairly impressive graphics, and some unique features out of games being hosted "in the cloud" ("Drivatars" in Forza 5, AI-augmented multiplayer in Titanfall). Its impossible to quantify this statement, but I think when all is said and done, this generation's sunset titles will seem, subjectively, 2-3 times "better" than launch titles; rather than better by 5-10 times, as in the past 3 generations.

We're coming back around to "sane" architectures this generation. In the past 3 generations there's always been at least one leading platform with an odd-ball architecture -- the PS3 with Cell, The PS2 with its Emotion Engine and not-quite-homogeneous dual-processors. and the Saturn with its literal symphony of processors. Stuff like that eats up R&D dollars, hurting saner competitors indirectly. Competitors were certainly different in generations before, too, but they were also simple, enough-alike, and essentially fixed-function platforms that there wasn't a lot of difference between launch and sunset titles, barring extra co-processor chips or add-on peripherals.

This generation we have two architectures who litterally share at least their main computational architecture -- the CPU and GPU core, and the amount of RAM -- with each other, and again with common PCs of today if we're talking slightly broader strokes. There's a minor difference of clock-speeds, a moderate difference in GPU clusters, and a philosophical difference in memory architecture, but that's far, far closer together than any two competitors have ever been, and nearer to contemporary workaday PCs than any time before. Also today, engine and middleware companies absorb a lot of that early R&D effort. All of that good juju adds up, and I just don't think that launch titles are leaving tons of untapped potential on the table.