The *only* useful thing that could come out of this is that bitcoin could finally jailbreak out of the 1 MB artificial lock-up, and segwit is in fact making this worse by trying to increase a very last time the amount of transactions per block, by delegating part of the block data outside of the "official" block size, but this cannot, as far as I understand, be extended yet another time.  In other words, segwit is the "end of the road" for more transactions if the jail of 1 MB blocks remains.  Yes, it will get us some temporary relief, but on the other hand, it has graved even more into stone the 1 MB limit and the "danger of hard forks" (which is a non-issue in most other crypto coins which do this regularly).

Once you break the jail once, you're free for ever.  However, if bitcoin doesn't succeed in overcoming this silly 1 MB limit now, it may even be much, much harder in the future to do so ; this is why it is of utmost importance that 2 MB blocks (with or without segwit) are adopted, to kill this 1 MB hard wall once and for all.

That doesn't solve fundamentally the essential design flaws in bitcoin, but at least, it doesn't turn a silly error into yet another extra fundamental design flaw for ever (I'm talking about the 1 MB limit).

As far as I understand it, the "BIP-91" technique is a technique invented to abuse of the "95% consensus soft fork rule" (BIP 9), and enforce it at any level, between 50% and 95%, without having to write explicit code for that and altering BIP 9, and hence, rendering BIP 9 moot, and using the "pure hash power majority" effect that imposes a soft fork on all miners if a hash rate majority adopts it.

Reminder: without any self-restraint, any soft fork that reaches 51% of hash rate can impose itself upon everyone, without inducing a chain split.  In order to render soft forks not so "dictatorship of the hash rate majority", but "full hash rate consensus", BIP 9 was introduced, that only activates proposed soft forks if they reach 95% consensus.

Of course, in reality, you better have some safety margin, because if your hash rate drops below 51%, you may see non-soft-fork blocks appear again in the chain (by the non-soft fork majority if they didn't adopt the soft fork or reverted back from it) which can be a huge mess.

BIP-91 style technique reduces BIP 9 back to potentially 51% majority over minority.

It goes as follows.  Pick your "safety margin for majority" (say, 80%, or 55%, or 75%...) A.   
Program the "real" soft fork under BIP 9, with 95% consensus needed to activate.

Propose a "meta" soft fork, by requiring signalling for the "meta soft fork".  The meta soft fork is simply a fork that doesn't do anything particular, except requiring signalling for the real soft fork.  If your meta soft fork reaches A, apply it (it is not difficult to apply it, contrary to the real soft fork which you leave to real devs: you only need to check on the signalling bit, and reject the block if it doesn't signal, that's all).

==> your A-level soft fork (not BIP9, but easily implemented) now requires that all blocks signal the real soft fork.  If A > 51%, this meta soft fork is IMPOSED upon all, so the surviving chain is the one with 100% signalling for the real soft fork, which now activates because 100% > 95%.

In reality, the real soft fork has been imposed with A-level majority, not with 95% consensus, but you made the software believe that you had 95% consensus, because you applied a non-BIP9 soft fork (the meta soft fork).

In other words, this kills the idea of BIP9, while keeping it still running.  Soft forks can now be imposed by any majority, even 51%, like soft forks do work.

Ok, so he said it, and he got that wrong.  It is not the only thing that he got wrong, but then, it is always easy to criticise with hindsight.  I think he didn't see the whole game-theoretical issue of the thing.  Like he didn't realize that the "hashcash" BRAKE on sybilling was totally reversed and became a STIMULANT once you REMUNERATE PoW.

Mathematically, you take out the LINEAR cost of hash cash with sybilling by proportional compensation, and you are left with higher-order corrections which are actually working the wrong way (economies of scale).

I don't know if and why he said that, but obviously it is wrong in the long term.  When the bloc reward becomes insignificant, ALL proof of work has to be paid by fees.  That cannot be "insignificant", because if it were, it would be easy to attack the chain with more PoW: it would only cost you more than the "total amount of insignificant contributions".

The principle itself of security by PoW comes from the fact that the total amount of value wasted (in the end, by the users) must be so significant, that nobody has any incentive to spend MORE to attack bitcoin.
This collective bill of bitcoin is paid by all the bitcoin holders, right now mostly due to inflationary pressure (inflation tax), but in the end, by those transacting.  In a certain way, this is *fairer*, because the cost of bitcoin's PoW is a service to those doing *transactions*, while with coin creation, all bitcoin OWNERS pay the bill (some kind of socialist redistribution: owners pay for the service to transactors) ; with fees, the transactors pay for themselves and the non-transacting owners are not affected (no inflationary pressure).

However, that bill is important, and for that bill to be important, transactions have to remain scarce.

So there is essentially NO WAY to make transactions cheap and fluid on a finally PoW protected chain without inflationary remuneration (as has been the case until now and in the decade or so that follows).

EVEN if you take transactions off-chain, they are STILL only secured by Proof of Work, and they will have to pay the bill IN ANY CASE or bitcoin becomes insecure.  No matter what.

This is simply because remunerated Proof of Work for consensus decisions was a BAD IDEA in a sound money coin.  (and sound money itself is an economically bad idea too, but that's another discussion: it turns the asset into a speculative asset falling into a deflationary spiral and makes it non-liquid as a currency).

Satoshi's claimed quote would make sense if the "cost of mining" were related to "the amount of data".  But it isn't.  The resource cost of data and networking is INSIGNIFICANT as compared to the ON PURPOSE wasting of resources in proof of work, which is not related to computing resource usage, but to CHAIN SECURITY.

As such  "the size of the transaction" doesn't matter AT ALL in the security model of bitcoin's proof of work.  It does, if the market squeezer is "block size".  But that's a silly squeezer.  As I said, the problem cannot be solved.

Yes.  (but... - see below)

The big invention of bitcoin is immutability of protocol.  (not of "software", of "protocol", that is "contract").  This is about the brightest thing that Satoshi did, but I think he didn't fully realize it (or for conspirational thinking, he knew it, but had his reasons).  Bitcoin's concept of a shared block chain that allows all users to come to consensus, and those not agreeing with consensus are simply ejected from the system, leads automatically to immutability if the decentralization is wide enough.  It is impossible to make ALL users agree upon an ECONOMIC change, because some will win, and others will lose, so the losers will never agree.  Of course, purely technical issues where everybody is convinced that it is better, is no problem, if there are no alternatives.  If there are technical alternative solutions to a technical problem, you will also get immutability because of "religious wars", people pushing for THEIR pet solution to the contract modification.

In other words, bitcoin's CONTRACT can only be modified if EVERYONE agrees over the single, same, obvious way to solve a difficulty in the advantage of *everybody*.  It is the sole exception to immutability ; in the same way that a normal contract can be modified if all parties agree upon modifying it.

Again, I'm not talking about the *software* that *implements* the contract (the protocol).  That software can evolve, and be as varied as one wants, in the same way that there are different e-mail clients, but they all talk SMTP.

However, in my opinion, Satoshi screwed up in several conceptual aspects of bitcoin, and these touch the very heart of bitcoin's contract.  These derive from the following "axioms":
- fake sound money doctrine: there should be a finite and fixed amount of bitcoins for t -> infinity
- remunerated proof of work will avoid sybilling of the consensus voting process.

If it weren't for these two fundamental mistakes, which are normally seen as the pillars of bitcoin, bitcoin would have been able to turn into a genuine currency generally used.  But because of this, bitcoin is deeply flawed in the long term.

A) the "proof of work" was meant to make it more and more expensive to "pretend to be many" (contrary to "firing up more and more full nodes", which Satoshi considered a danger if they had to vote, because you can easily Sybil it).  But you KILL ENTIRELY this "making pretending to be many" expensive if you REMUNERATE "pretending to be many" (using a lot of hash power).  --> this is the root cause of bitcoin's unavoidable centralization: if you *remunerate* the consensus decision process, and there are economies of scale (unavoidable), then you will have the OPPOSITE effect than decentralization: you will push to a power law distribution of the power to vote the consensus protocol, and hence end up with a few tens of deciders (the pools) at best.

But the problem is that because proof of work is in fact "proof of economic waste", you HAVE to remunerate the consensus decision or nobody is going to make it, leaving the system insecure.

B) and this brings us to the caveat (my "but - see below"): combined with sound money doctrine, you will have to diminish bloc rewards more and more.  In the end, the remuneration of the proof of work will have to come from FEES.  In order for fees to be sufficiently lucrative to allow sufficient proof of work, TRANSACTIONS NEED TO BECOME SCARCE.

So this is the "but".  If you allow unlimited blocs, there's no way to remunerate the centralized deciders of bitcoin in the long run.  Satoshi simply *didn't know how to solve that* and *there is no solution for his problem* in the frame of the axioms he put forward.  Maybe that's why he introduced the 1 MB limit.  Because he didn't know how to solve his unsolvable problem.

It is the information that is essential if you want to trace the origin of a transaction (in other words, if you want to deanonymize the payer).  If you run a full node, it might be *you* who is creating this transaction.  Or it might be a light wallet connected to your full node, at which point, looking at your internet connections around the time of the relay may indicate the IP number of the computer that was running the light wallet, indicating who might be the payer.

This is in fact one of the biggest advantages of running a full node: plausible deniability that you are the one sending out a transaction.  (but you can also use a light wallet and, say, tor or a vpn).

This.  A big amen.  The essence of crypto is here.  A zero sum game, a pump of value from greater fools to earlier fools and those having their business around it (like miners and exchanges).  Which doesn't mean it cannot be played, if you think that there are still a lot of greater fools, that will pump their value your way, of course

As you point out, it is not even a zero-sum game, because mining makes it lossy.  Value has to be destroyed continuously by bitcoin and every other PoW coin, in order for it to keep on running.   As long as this was compensating seigniorage, this could have a good effect (creating coins wouldn't be pure seigniorage as you have to waste most of it, the inflation tax doesn't go into anyone's pocket (= seigniorage) but is burned into heat of mining equipment).  But in the long run this is not sustainable: you have a zero-sum asset with a value leak in it.

My point was that if the decentralized system is well-designed, then entities meeting secretly/publicly under whatever conditions shouldn't matter because they wouldn't be able to do anything about the system.  If one is afraid that the "rules" laid out in certain meetings are "unfair" or dangerous, it means that one thinks that whatever happens in those meetings may have an influence on the system at hand.  But that simply means that the system's decentralization (and hence immutability) failed.
If the system were well-designed, no meeting or no collusion could ever have any influence on it, and all those meetings would be lost efforts.

Now, it might very well be that bitcoin HAS a stronger form of decentralization (and hence immutability) than we think, and that all these meetings ARE in vain.  That would be interesting to see.  However, then bitcoin faces another problem: the fact that its current (immutable) protocol is driving it into a wall with the block size limit.

So bitcoin is damned if such meetings can modify it, and is damned if they can't.

You are perfectly right.

The real issue is that one wants to push agendas, and uses this false problem to do so.

While it is true that making bigger blocks makes a bigger block chain that is not to be afforded to be downloaded by the most modest of potential bitcoin users, one forgets to mention:

1) that to be a bitcoin user, one doesn't need the block chain in its whole, and a light client is perfect.
2) that the number of full nodes that do not mine has nothing to do with the decentralization of bitcoin.

The nasty thing about bitcoin is that it has lost most of its ideal decentralization, but not because of the size of block chains, but rather because of the economies of scale of proof of work.  Bitcoin's design was exactly made such that "full nodes" have nothing to say, and only MINERS decide on consensus (of chain history and protocol).  Only proof of work is a "credential" in bitcoin, exactly because Satoshi wanted to avoid the number of full nodes to play any role (it is too easy to sybil it).

And the mining entities, that make the decisions, are the mining pools and there are essentially 20 of them.  That's the real state of "decentralization" of bitcoin (the majority is even in the hands of only 5 of them which may be sock puppets of 1 or 2 people).

In order to hide that ugly face, a lot of wind is made concerning the importance of "Joe with his full node in his basement",  but that node has nothing to say.  Joe's node can only download whatever single block chain the miners mine, or stop.  If Joe's node stops, it is as if Joe switched off his node.

A light client can check that one isn't serving bullshit by just downloading the header chain, which contains all the proof of work ; and once that header chain is verified, any specific transaction can be checked by verifying the Merkel tree branch in the block at hand.  A light client doesn't have to check for double spends, because the miners already did so, and if the miners decided amongst themselves to include a double spend in the consensus and continue to keep consensus over that (continue to build blocks on top of that), then in any case, you cannot do anything about it, and that double spend is now in any case part of bitcoin.

So the raw truth of bitcoin is that as a user, you are in any case at the mercy of what the miner pool consortium decides (it has full consensus decision power), whether that is "according to the rules of bitcoin you thought were in place or not", and once you accept those decisions (and you can't do anything about it), a light client is sufficient to check whether a given transaction has been decided by that pool consortium or not (nobody can fake it).

There are some advantages to running a full node, and one has to find out for oneself whether those advantages are sufficient to spend the needed modest investment of such a node is worth it or not.  But it has nothing to do with the "decentralization" of bitcoin.  Because bitcoin was exactly designed for it not to matter.

That ugly truth cannot be heard, and that's why a lot of fuzz is made by bitcoin evangelists about the importance of full nodes.  But they are about as useful for the decentralization of bitcoin, as hanging a talisman around your neck is against car accidents.

This illustrates the failed decentralization of bitcoin.  One shouldn't "complain" that people are having secret gatherings.  If those secret gatherings have any potential EFFECT, it means that bitcoin was prone to collusion (and hence wasn't decentralized) in the first place.

Complaining about this is a bit like complaining that you have a cryptographic protocol, but that there are people having secret meetings about how they can crack your cryptographic system.  If you are to be afraid about people trying secretly to crack your system, then it means that your crypto system is worthless.  If a system claimed to be decentralized has to fear the "secret gathering of a few hundreds of individuals", then it is not decentralized.

Just as with cryptography, one should say: "try to collude as much as you can, try to modify bitcoin to your hand, the more you do so, and the more you fail, the more you prove that our concept of our system is working correctly, because it was designed to resist exactly that".

The big failure of bitcoin is that its protocol is made such that it NEEDS to be modified, which can only be obtained by breaking its resistance to power (that is, the ability of some to change its rules).  So bitcoin will fail technically (because of block limit) unless it fails conceptually (because centralized and power structure).

I think this is wrong already.  It is based upon the belief that bitcoin's decentralization (that is, the spread of the powers of decision in bitcoin over a lot of non-colluding entities) has something to do with the amount of copies that are kept around the world on different storage apparatus (Joe's node in his basement). 

It is indeed not feasible to spread 2 GB blocks every 10 minutes to EVERY JOE node in his basement, because there are many Joe that won't invest a few $1 000 in the necessary equipment.  As such, your statement is right.  But the fundamental mis-understanding here, is that Joe's copy of the block chain matters AT ALL in the decentralization of bitcoin's decisions.

Only 20 entities make all those decisions: the mining pools.   These entities are the sole entities being able to even MAKE a right block chain.  It only takes the chain of HEADERS to VERIFY that the chain of headers you have, is the "best block chain" and even to make a failed imitation of it would cost a fortune and would easily be detected.

So in reality:

1)only about 20 entities BUILD the sole block chain out there
2) every smart phone in the world can easily check that he has the right header chain downloaded, made by these 20 entities (can check the proof of work in that header chain).

Once you realize that, you also realize that the only entities that REALLY NEED to have this full chain with 2 GB blocks, are those 20 entities ; but of course, anyone willing to invest enough in it, can have that chain too - only, apart from big users like exchanges, THERE IS NO POINT in having this amount of data:
a) possessing it, and transmitting it, doesn't change ZILCH to the decision process in bitcoin (whether you have these data or not, doesn't alter the decisions that the sole authors of the block chain, the 20 entities, make)
b) you can verify for yourself that the sole true chain out there corresponds to the headers which ARE very small and independent of the block size
c) those needing it, are so much invested in bitcoin in any way, that for THEM there is no problem in having these data transmitted and stored quickly and smoothly.  (exchanges and mining pools DO NOT HAVE technical problems with such volumes, not more so than any serious data center has in serving such kinds of volumes which are small to other internet usage).

==>  there is strictly no technical problem for the involved entities to process so much data, even as of today, and all the others DON'T NEED IT and it doesn't serve any purpose of "decentralization"to have your own copy of the block chain - but if you WANT to you still can.  For reasons of privacy, or because you want to do chain analysis and try to break other people's privacy.

The "waste of storage resources" for those few entities serving the block chain (because they are also the sole entities MAKING the block chain, and taking all decisions on it) is ridiculously small as compared to the inherent waste of resources imposed by proof of work, wasted by those same entities.  In other words, given all the wasting that goes on in bitcoin, the "block chain size" issue is microscopic and irrelevant.
This whole debate is meaningless ; the blocks can have just any size, only a few copies need to be available to users and we're talking about amounts of data far inferior than most middle sized data centers are already handling with a smile.  There's nothing "decentralized" in working as a distributed proxy in your basement for the sole chain that is out there and on which you have, in any case, nothing to say.


It is fairly obvious that if you run LN nodes for profit (and you have to, because otherwise it will be lossy with all the settlement fees), the economies of scale scale strongly with committed capital, making big hubs able to offer much more competitive LN channels to customers than small users. 

I'm usually writing stuff that is perfectly logical, but that hurts the religious feelings of indoctrinated people that cannot think outside of their convictions.
Visibly a fundamental such belief is for instance, that full nodes help anything in the decentralization of the power structure of bitcoin, while bitcoin's protocol was EXPLICITLY designed NOT to give any such power to full nodes, and that (contrary to other aspects of bitcoin's design) that part of the design actually reaches its goal: full nodes have NO decision power.

From that ill-informed religious dogma, follow a lot of contradictions, which I patiently illustrate.  These obviously logical conclusions are then provoking strong mental resistance by the True Believers, and I have to say that I take some pleasure in poking logical holes in that belief, each time I find another illustration of it.  The idea of minimum block size, absurd as it may sound at first sight, but brilliant as a very secure and easy implementation of a guaranteed bilateral hard fork (which was for instance the problem with BU, that was unilateral) is again hitting such a dogmatic opposition, while there isn't any to be had.

Apart from big users (such as exchanges) and of course, mining pools, *nobody needs to store, transmit, or verify* the block chain, so if these entities can afford the resources to handle "a few DVD a week", bitcoin will work, and as their investments in bitcoin are WAY beyond the cost of "handling a few DVD a week", there's no problem.  A dedicated Joe can also have his "a few DVD a week" in his basement, but he only does that *for his own fun*.

This is hurting of course the fundamental belief system of bitcoin, and one "shouldn't waste computing and network resources because poor Joe and his old second hand Dell desktop on his 56K modem will not handle this and Joe is important in bitcoin".  It isn't.  Joe's second-hand full node has nothing to say.  But this cannot be heard by true bitcoin believers even though it was the core of bitcoin's design.   Bitcoin is now built on "wasting resources" where almost a billion dollar is wasted a year, but one is discussing about the craziness of copying a few DVD a week on those few data centres that are the essence of bitcoin, and do take, in any case, all decisions.

When you read such stuff, you know how far gone people are in deluded group think.  And yes, it is well known that people putting logical arguments on the table that hurt the fundamental dogma's of group think are badly considered by the indoctrinated.  That's what religious wars are all about.

In fact, I think that the LN is a good way of adapting to what crypto REALLY looks like:

big exchanges with customers trading on it

LN is quite well made for that, and LN links between customers and their exchange(s) is a perfect way to make the link between the customer and the exchange honest and trustless.  By using a LN channel to your favorite exchange, you don't give then coins in return for web-site IOU, but you are still master of your coins, the exchange cannot apply fractional reserve banking, and cannot put up messages like "withdrawals impossible now" and things like that.

LN is perfect for the relationship exchange-customer (which is what bitcoin and most crypto *essentially* is).  Moreover, LN would then even allow links between exchanges.  They would be the central hubs in the LN network.  So through this network, you could pay other people that have accounts on other exchanges, and this would be much more trustless than now, when you have given up your coins, and you hope that the exchanges will treat the IOU correctly (that is, that they will honour in the end, your supposed right of withdrawal).

But this is about the only way of using LN: making exchanges into big banks.  Which is, what crypto is in any case, about for 99%.

Marketcap in an illiquid market can be bogus.  But market cap in a liquid market is a relatively good estimation.  If daily volume is $3 billion or so, and the market cap has fallen by 20 billion over 5 days, then there HAS essentially been going 20 billion gone out of the circuit give or take a factor of 2.

Well, once it is out of the bottle, whether somewhere implemented in one or other toy version of bitcoin doesn't matter: the idea of MINIMUM block size to impose a clean bilateral hard fork over block size is a very smart idea.  Whether it is done by segwit2x or not, doesn't really matter now.  They SHOULD do it.

(however, if gmaxwell tells us that it "rejects its own blocks" because they are smaller than 1 MB, one would think that it DOES implement a lower block size limit).

In any case, the thing with block size limits is a fundamental design problem in bitcoin, due to its badly designed economic model, and its centralizing compensated PoW game theory.  All these discussions are just small fixes on what cannot be fixed fundamentally.

Of course I'm OK with that.  Only the miner pool nodes and a few big players need to keep an integral copy, and for them, such an investment is not a problem.  All normal users can use light wallets. 

In fact, "when the blocks are full" and an upper bound is hindering the system, changing this former upper bound against which the demand is hitting, by a lower bound, is not going to waste much in the hypothesis of continued adoption.  If 1 MB blocks are most of the time full (they are), then requiring a 1 MB LOWER bound will only waste those few differences of those few smaller blocks than 1 MB (which are almost not there, otherwise the 1 MB limit wouldn't be a problem).

So the "wasted space" is limited by changing the former upper limit which became to small and should normally naturally be crossed by changing it into a lower bound.  But the nice thing is that the old and the new blocks are now fully incompatible, which is what is needed for a clean hard fork (bilateral incompatibility).


As I said, no, because miners can fill them up artificially.  They won't be letting their expensive mining hardware sit idle 30% of the time, waiting for people to send transactions before they can START mining.  No, like now, from the moment there's a block found by a competitor on which they will build, they will make a legal block, with enough junk in it for it to be a legal >1MB block so that they are not wasting time NOT mining.


No, not at all, because miners can put as much junk into a block as they wish.  They can make many fake transactions, even without fee, or by paying a fee to themselves.  Miners are never short of transactions if that's what they need.
They prefer of course juicy real transactions with nice big fees from other people, but if they just need to generate random transactions at no cost to be able to mine, they can always do so, unlimited.

Well, it is a very smart way to make in a very very simple way, a bilateral hard fork, which is a GOOD thing.  And BTW, nobody cars about "block chain bloat".  Mining pools have enough hardware to do so, and normal users only need light wallets.  Those really wanting to play with a full node in their basement will not see much difference with today, where the average block size is already close to 1 MB.

I never thought about this simple way of making a bilateral hard fork.  Usually, one thinks about a modification in some or other format or so.  But obliging >1MB is genius and simple. 

Because bullshit.  
Segwit contains some relatively good ideas, but they are not necessary.  Bitcoin contains conceptual and technical choices which are less than optimal, and segwit can improve somewhat on them.

So, somehow, segwit is some technical improvement of some aspects of bitcoin technicalities for more sophisticated uses of bitcoin that aren't really used much today.  The MAIN reason for segwit is that core wants to push changes, and wants to push the lightning off-chain payment system.    Core has been selling a lot of bogus reasons of why segwit is the solution to scaling.

The two main bogus reasons are:
- we cannot afford hard forks, that would be too dangerous, and simply changing *one single parameter* in the existing software, namely the block size limit, is technically a hard fork.  Hard forks are not dangerous if they are consensual, ethereum has done many of them, monero has done many of them.  A hard fork where a single parameter is simply changed wouldn't have been a great deal, but core made a great deal out of it, to be able to push segwit as a sole solution to the silly block restriction.  They do this by allowing (in segwit) "out-of-accounted-blocksize data" (the so-called witness data), and hence allow more ACTUAL data per block than is accounted for (there's only a part of the actual data that goes "into the block", but that block is depending on out-of-block data that gives the full story of the transactions).

- we need small blocks so that there are many voluntary full nodes that "keep bitcoin decentralized".  That's probably the biggest bogus reason.  Full nodes don't decentralize anything as they have no voting power in any consensus decision (on purpose replaced by proof of work by Satoshi, to *avoid* possible sybil attacks on the number of full nodes running and voting). Full nodes only INFORM their owners. But moreover, a true full node would need just as much network and storage and computing effort to verify the "out-of-block data" too, as if these data were included into the block.  Having a block of 1 MB (so, respecting the official limit and no hard fork), with next to that, 3 MB of "out of block" transaction data related to the "fingerprint" in the block, is just as cumbersome to process than old-fashioned 4 MB blocks: the amount of data transiting and processing is the same.  The excuse "but you don't NEED to verify this out-of-block data if you take the fingerprinting for granted", is exactly the same reason that one can use to use a LIGHT WALLET, where you don't NEED the full blocks, but just the chain of headers, and the branches of the Merkle trees that concern your transactions.

Finally, once segwit has augmented the effective block size to 4 MB even though it makes believe old nodes that it doesn't need more than 1 MB and hence is not technically a hard fork (old nodes don't understand zilch of segwit, but accept the ununderstandable blocks because they look like funny but legal transactions to them), there is no further block size increase possible EVER without, finally, the dreaded hard fork of changing a single parameter.

So 20 years from now, bitcoin will STILL run on a room of 4 MB per 10 minutes even if at that point, 3 GB blocks would not be any problem.  These 4 MB every 10 minutes, which is all you can ever extract out of soft forks in bitcoin (which is what segwit is doing), will by FAR not be sufficient to be able to guarantee any settlements of a large Lightning Network that would run on top of segwit, if ever.

In other words, the soft-forking dogma will restrict bitcoin for ever, and doesn't solve any long-term scaling issues.  It will even render off-chain lightning network applications DANGEROUS if ever they pick on because settlement cannot be guaranteed.

Without the soft forking dogma, just increasing the block size would have solved the issues since long ; there's nothing to be won to not do this, whether one wants segwit for technical reasons or not, whether one would like to experiment with off-chain payments or not.  In any case, this has no influence on the decentralization of the consensus decision within bitcoin, which is as of now, pure proof of work, and is not much decentralized already.

Well, I don't see why a miner wouldn't fill up the block to 1 MB with dummy transactions, just to not waste his hash rate, and get the next block while all others are waiting to start to mine until they can fill up a 1 MB block...

A lower limit on block size will never be a problem, because miners can make blocks as large as they want with own dummy transactions.  This makes that miners cannot play games with small blocks as they are now not part of the protocol any more, and all "relative network advantage" between 1 MB and 2 MB is much smaller than between 10 KB and 1 MB.

Alt coins matured, and are now somewhat independent of bitcoin.  You can see this by the fact that since quite a while now, bitcoin's volume is 1/3 or even less of total volume, which means that bitcoin is NOT the "partner" in most alt coin trading (if it were, bitcoin's volume could never be less than half of the total volume, and in fact more, because there would be bitcoin/fiat + bitcoin/altcoin trading).

Alt coins are now separate from bitcoin.  But they are all part of the crypto speculation domain.  Crypto has been in a huge bubble, since, well, since it got into existence essentially :-), but it got crazy since beginning of 2017.  No surprise, crypto IS speculation for most part.  (the "real world use" is negligible as compared to speculative trading, also for bitcoin).

It is true that price-wise, altcoins are falling faster than bitcoin (as they rose faster during the rise of the 2017 bubble too), so bitcoin is gaining market share (from 37% at its lowest to near 50% again).  But this is not because people are fleeing from alts into bitcoin.  It is because people are fleeing somewhat faster from alts than they are fleeing from bitcoin.

There is no distinction of principle (any more) between bitcoin and alt coins.  Bitcoin's first mover advantage is eroding away (and is seemingly "winning" again somewhat, but simply because it is falling somewhat slower, not because it is taking in from what came from the alt scene).

Hard to say what will happen next in this entirely speculative game.