How "decentralized" is bitcoin - and how transparent is its "consensus" finding?

So my question is simple: Who is actually deciding about changes in the bitcoin protocol, how was it intended - and what does that mean in the long run for public PoW based blockchains?

My understanding

In the beginning ... there was ... paper - well - Satoshi Nakamoto's White Paper.
In this paper, the mysterious Satoshi Nakamoto (could be male, female, a group of males, females, mixed, Asian, European, Indian, South American...you name it; to make it easy I will just refer to "him" as one male person; call me simple that way) wrote in the introduction section (section 1.) of that document:
Quote: using a peer-to-peer distributed timestamp server to generate computational proof of the chronological order of transactions. The system is secure as long as honest nodes collectively control more CPU power than any cooperating group of attacker nodes.

Note, that in this document there are words, you will not find or are rarely used, yet, they seem very important today when looking at public PoW blockchains like bitcoin:

For example, Satoshi uses the word "miners" only once and in an anology: "... constant of amount of new coins is analogous to gold miners ..."
This is the case, because in this paper there was no distinction between nodes and miners - both were the same.
Actually maybe its important to recall this - the basic idea was that everyone using the system was also a full node. Full stop.

No exceptions in this White Paper, called "Bitcoin: A Peer-to-Peer Electronic Cash System" (that is funny, too - the "Cash System" dimension...but well another time...).

In contrast to the White paper, today we distinguish between miners, full-nodes, wallets and light wallets...well actually there does not seem to be a clear distinguishing definition for "wallet" - it can be a full node, but I could not find any "paper", that demands a wallet to be a full note - yet "light wallet" suggests exactly that...well...

I personally assume, that not many of the new user of bitcoin know the differences, anyway...and what this actually means for their part in influencing consensus finding thus where the blockchain is headed...

Lets take the current topic of bitcoin as an example: Shall bitcoin achieve scaling abilities using a solution called "Segregated witness" (SegWit) or "Bitcoin Unlimited"?

As far as I understand, a change in the protocol, may it be one or the other, would have to involve a majority of the miners "signalling" their approval. "Signalling" hear means, they add a corresponding piece of code/value to their mined blocks.

Some bitcoiners (early or medium early adopters) surprisingly also look at nodes signalling - as if this was of importance?
E.g. take this tweet by Tone Vays as an example:

In contrast to that, the site coin.dance shows a complete different picture as it claims looking at signals from miners:

So either my assumption is wrong, or even old bitcoiners can get confused using terms like "node", "miners" and "wallets"?

IMHO there is a new (?) danger in this "node signalling", though:

New users to bitcoin mostly get directed to the Bitcoin Core wallet as "bitcoin.org" is a major entry point into this blockchain - and on "bitcoin.org/download" the Core software is offered.
So like Tone Vays shows in his image - the software called Bitcoin Core signals approval of SegWit from versions of that software greater than 0.13.2.

So if you use that "wallet" - well - you are making a political statement regarding the scaling issue of bitcoin.

Most older bitcoiners I know, actually do not see an issue here - they say things like "new users don't use full nodes"...or even worse "leave-me-alone"-kind of replies ;-)

One interesting response was, that while (non mining) full nodes played no role in consensus (=bitcoin protocol defined majority for change to protocol) finding, they did so naturally in relaying miners blocks to the complete network. That was enlightening in itself as the "relaying of information" is also a big issue - even if not talked about as often as "mining".

And using stats that show the "not mining full node" support for one of the two solutions like Tone Vays did clearly could be seen as a bit manipulating, as an unknown amount of the humans running those nodes have no clue they are showing support in that context.

What I feel kind of critical is that the actual information on that Bitcoin Core software update only states the segwit signalling in a kind of hidden statement 2 complete scrolls down that page (you'd think, it would be an important information needed at the very top!).
At the top that page simply says:
Quote: This is a new minor version release, including various bugfixes and performance improvements, as well as updated translations.

I find that a poorly managed communication - at best!

So what does that mean generally for public PoW blockchains?

Well, for one - there should be clear and easily retrievable sources for term definitions, such as "node", "miner" and "wallet".

And then? - What about this consensus

Well, to be honest, I think, the purely technical solution of finding consensus in a decentralized network - is not the end of the road.
Projects like Ethereum (while clearly with its own issues in the field of decentralization and mining) are using a lot of its ressources to solve PoW isseus - either by enhancing PoW or by switching to PoS (Proof-of-stake).
Others, like DASH use a kind of mix between PoW and PoS...but just like Ethereum, DASH again has its own issues to solve as well.

So at the end, I can only leave you with this understanding: Decision making in regards to the bitcoin protocol are NOT transparent and NOT free of (maybe not intended) manipulation. Be aware when you read any "stats or news" on this topic!