New video: Why the blocksize limit keeps Bitcoin free and decentralized

[tvbcof]

Video: http://www.youtube.com/watch?v=cZp7UGgBR0I

Website: http://keepbitcoinfree.org/

Pretty simple right now, but this is the beginning. For those of you at the 2013 conference, I'll be giving a presentation about off-chain transactions on Saturday as part of the tech stream.

I just got around to having a peek at the web site now that it has some content.  Cool!

I implore you to 's/stenography/steganography/g' on both the site and in IRC

  http://en.wikipedia.org/wiki/Steganography

---

I might clarify that my interest in Bitcoin and my interest in this aspect of Bitcoin is NOT really a reflection of my dis-satisfaction with our mainstream economic systems today.  Unlike many here, I think these are actually working surprisingly well.  My concerns are very much focused on what the landscape looks like tomorrow (figuratively speaking.)  And it is highly political to me.

Distributed crypto-currencies have a very real potential to be disruptive to mainsteam economic systems, and these systems are of extreme value politically.  We're not playing tiddly-winks here.  To be dismissive of the potential for a very significant backlash is unbelievably naive to me.  I would be completely amazed if 'the powers that be' rolled over and played dead as distributed crypto-currency technologies advance unless they have simply no other realistic alternative.  In other word, I expect every chink in the armor to be mercilessly explored.

[adamstgBit]

idk if i like the idea of having to store bitcoin in centralized payment processors aka bitcoin BANK

[jubalix]

music very repetitive gets annoying

hang on why does it become un profitable for miners when say the block size is doubled?

Also as BTC goes up by more usage, then fees become a smaller as a fraction of the BTC and so stay the same in buying power for power and hardware.

what am I missing here?

[adamstgBit]

idk if i like the idea of having to store bitcoin in centralized payment processors aka bitcoin BANK

i guess, your already doing that when you send BTC to a website like https://www.havelockinvestments.com/ and buy and sell their bitcoin stocks....

[adamstgBit]

music very repetitive gets annoying

hang on why does it become unprofitable for miners when say the block size is doubled?

Also as BTC goes up by more usage, then fees become a smaller as a fraction of the BTC and so stay the same in buying power for power and hardware.

what am I missing here?

their theory is that mining will require too much bandwidth if the block size is too high.

pretty soon you'll need a direct connection to the backbone to be able to download the blockchain fast enough, to be able mine. ( or somthing like that? )

this is make it impossible to be an anonymous miner in the network, very few people will be able to be miners, b4 you know it only a handful of miners have control over the network.


their idea is in the right direction, it would be cool if these payment processors could be built in such a way that they do not require trust.

some kind of automated bitcoin payment processing system that is open source and secure.
 not sure if possible... If not you'll need to trust your coins to a central entity to do these off chain TX

[jubalix]

music very repetitive gets annoying

hang on why does it become unprofitable for miners when say the block size is doubled?

Also as BTC goes up by more usage, then fees become a smaller as a fraction of the BTC and so stay the same in buying power for power and hardware.

what am I missing here?

their theory is that mining will require too much bandwidth if the block size is too high.

pretty soon you'll need a direct connection to the backbone to be able to download the blockchain fast enough, to be able mine. ( or somthing like that? )

this is make it impossible to be an anonymous miner in the network, very few people will be able to be miners, b4 you know it only a handful of miners have control over the network.


their idea is in the right direction, it would be cool if these payment processors could be built in such a way that they do not require trust.

some kind of automated bitcoin payment processing system that is open source and secure.
 not sure if possible... If not you'll need to trust your coins to a central entity to do these off chain TX

these seem like technical limitations that should be overcome given time and moores law, eg a120 GB a month internet account is not that much these days...and the speed is pretty good...many times faster than it used to be

there may be some innovations in the software that help as well

I am unconvinced about the central premise

[adamstgBit]

music very repetitive gets annoying

hang on why does it become unprofitable for miners when say the block size is doubled?

Also as BTC goes up by more usage, then fees become a smaller as a fraction of the BTC and so stay the same in buying power for power and hardware.

what am I missing here?

their theory is that mining will require too much bandwidth if the block size is too high.

pretty soon you'll need a direct connection to the backbone to be able to download the blockchain fast enough, to be able mine. ( or somthing like that? )

this is make it impossible to be an anonymous miner in the network, very few people will be able to be miners, b4 you know it only a handful of miners have control over the network.


their idea is in the right direction, it would be cool if these payment processors could be built in such a way that they do not require trust.

some kind of automated bitcoin payment processing system that is open source and secure.
 not sure if possible... If not you'll need to trust your coins to a central entity to do these off chain TX

these seem like technical limitations that should be overcome given time and moores law, eg a120 GB a month internet account is not that much these days...and the speed is pretty good...many times faster than it used to be

there may be some innovations in the software that help as well

I am unconvinced about the central premise

the central premise is true, 1MB is the just about the limit for mining bitcoin under TOR.
not sure why we need to have minners run under TOR tho... its not like mining bitcoin is illegal

in any case their idea of a central processor is good for another reason, it would solve the need to wait for 6 confirmations problem.

[Daily Anarchist]

Someday people are going to look at 1 MB like they do a 1 KB. Keeping it at that arbitrary size doesn't help anybody. I'm sure there is a decent algorithm that will allow for an increasing sized block without it ruining bitcoin's decentralization.

Furthermore, I still don't understand why full nodes don't receive some compensation? Why only miners?

[amincd]

I see the off-chain solutions as being every bit as ad-hoc as the transfer nodes and miners (if need be.)  Every time one gets a mallet on the head, three more pop up out of a gopher-hole somewhere else in the world.  That, my friend, is resilience.

To explain why I think relying on third party payment processors is what would end up making the BTC-economy look like the modern banking system, and why I think it would be very un-adhoc-like if these third party payment processors were to handle global-scale transaction volume, I'll explain how I imagine it will work:

You have a merchant who wants to accept BTC payments. He can't accept them through the Bitcoin network, because transaction fees are $20, so he needs a payment processor. This isn't a removed payment processor like BitPay that simply receives BTC payments from customers and converts it to fiat for the merchant the same day.

The payment processor would need to accept BTC-credit from a customer, and then hold it in an account on behalf of the merchant or sell it to a party that would be willing to buy that BTC-credit, and deposit fiat in the merchant's bank account.

Accepting BTC-credit is not a simple thing. It means having to be able to trust the BTC-bank that the customer has BTC-credit with. The payment processor won't just trust any fly-by-night BTC-bank, since it hasn't proven itself trustworthy.

More likely, there will be a network of trusted BTC-banks, that have trust-based relationships with each other, and use a centralized clearing house, like modern banks do, to settle their debts at the end of each day with on-chain transactions.

Parties who can't join this centralized clearing house, because of their size, or regulations, or any number of reasons that people are currently locked out of the modern banking system, are effectively locked out of the BTC-economy.

Merchants can't accept their credit, because the payment processor they use doesn't accept the credit of any BTC-bank that doesn't participate in the same payment clearing network as them.

The competitive advantage of the parties that have large user bases, or are members of networks with a large number of members, would be too great for outside competitors to overcome, leading to a very centralized and static BTC-economy. It would be just like Visa/Mastercard, or the bank payment systems, like CHIPS.

Regarding the current banking system being a "fully automated p2p solution", this is not true at all. The current banking system is based on contractual relationships that require trust and regulatory access. A fully automated p2p solution has no contracts and every thing is done through protocol.

If a node in a p2p network goes down, no one loses their money, because it's a distributed p2p network where there are thousands of other nodes that store the same information. They communicate through the internet, and form a global network. Setting up a node is only a technical process, with no trust-based relationships required. As long as you have the bandwidth and hard drive space, you are equal in standing to every other node.

I don't believe that things would play out as you imagine as long as the barrier to entry for being a payment processor is low (which, of course, facilitates both competition and the ability for a processor to take bigger risks as they have less to lose.)

The OP has a conception of payment processors being able to demonstrate proof of stake in backing store ownership.  I like this, but also things could work well with 'off chain' meaning being more completely de-coupled or trust-based.

The barrier to become a payment processor that has trust-based relationships with other BTC-banks would be much higher than the barrier to become a Node.

We see in the real world that financial systems built on trust become highly centralized, for reasons independent of regulations, like, well, payment processors, and the way people rely only on the largest networks (Visa/Mastercard/ACH).

This is because trust takes time to build. Centralized BTC-banks are also much more attractive to government as targets for regulatory action, because a takedown of one causes a disruption to the activity it's involved in and is not easily repaired (trust takes time to build).

Nodes on the other hand, because they are only one of thousands of nodes storing and propagating the same data, can be taken down without any disruption to the bitcoin economy. A node can also easily pop up to replace any that is taken down. It's only a matter of mustering the technical resources because no one has to be able to trust a node operator for the node to be immediately plugged into the p2p network. Hashes and digital signatures don't lie.

I seem to have a little more confidence in the potential for crypto-currencies to de-couple themselves from fiat than do you.

I'm not quite sure what you mean here. In which way have I shown little faith in de-coupling from fiat?

I actually have no huge problem with 'trust based' relationships at some levels in a system.  These tend to increase efficiency and work well for long enough durations that they are plenty useful for exchange functions.  

I don't have a problem with trust-based relationships in the BTC-economy as long as transaction fees are low enough so that people can very easily use the p2p network instead of third party BTC-banks, so that they have an alternative to trust-based networks.

Where they bother me is when I am thinking about a need for stored value decades out.

We we can agree to disagree here. I think it's quite important to not have to rely exclusively on trust-based relationships for the main payment channels of bitcoin.

To me, Bitcoin is just the right size when off-chain processors can use it as a rock solid system for periodically squaring accounts, and most users can use it for a 'retirement' value store.

Again, we can agree to disagree. I think the value of bitcoin will be threatened by competition if its network transaction fees are high, which would threaten its role as a retirement value store.

[Peter Todd]

jdillon, tbvconf and others: Thank you!

FWIW As it stands at I've gotten about $5750 worth of donations for the video, including the initial $3000 committed prior to making it, so I'm only about $1250 away from breaking even. I knew going into the project that I was taking a big risk, but in the end it's all worked out.

Again, thank you!

[Daily Anarchist]

jdillon, tbvconf and others: Thank you!

FWIW As it stands at I've gotten about $5750 worth of donations for the video, including the initial $3000 committed prior to making it, so I'm only about $1250 away from breaking even. I knew going into the project that I was taking a big risk, but in the end it's all worked out.

Again, thank you!

Despite my disagreements with the video I thought it was really well done. I hope somebody does something as equally good for the opposition.

[datz]

I like this but I do not see where you fix the transaction fee problem. The video states that transaction fees will continuously rise due to small block size and increased adoption. So, you are basically advocating transaction fees in order to keep Bitcoin decentralized? This is where Ripple shines - no mining necessary. We need to reexamine other protocols as a community and stop trashing technological progress.

[kodo]

Wow i learned a whole lot reading this whole thread!

[d'aniel]

If anyone's interested in the anonymous mining issue, here's an informative exchange found in this thread: https://bitcointalk.org/index.php?topic=157141.0

I think you are massively over-estimating the difficulty of mining anonymously (as is usual with these debates).

Firstly, there is no particular reason mining on a pool requires a lot of bandwidth. Stratum with high difficulty shares already cut bandwidth usage very low.

For running the full node/pool itself, resource requirements are low. You can connect to the P2P network to gather transactions just like any other client. Nobody knows you are mining, even without Tor, as your behaviour is indistinguishable from any other node. Nodes can synchronise their mempools are startup and then know what each peers preferred block contents is likely to be: once solved, a block can be transmitted as a delta against the expected contents. This has the side effect of increasing bandwidth requirements for people who want to mine empty blocks, which is satisfying.

However, even without that change, I can't see a time when mining anonymously is impossible. Your argument, as always, relies on the assumption that "things" cannot possibly scale up, where the thing here is Tor. Tor can scale, so even if mining ends up requiring a lot of bandwidth it doesn't change anything.

Mike, lets suppose for a second that I am correct, and Tor bandwidth and other anonymity technologies do not scale with the bandwidth possible at VPN/co-location providers, and thus running mining (not hashing) operations is not possible anonymously.

What do you expect to happen?

Not much?

In the event that some jurisdictions with aggressive regulators try to impede mining, it will migrate elsewhere. Only if all mining is made illegal everywhere would it be a problem, and that's equivalent to a worldwide ban on Bitcoin, at which point it doesn't matter anymore.

Quote from: Mike Hearn on April 02, 2013, 10:05:44 PM
Nobody knows you are mining, even without Tor, as your behaviour is indistinguishable from any other node.

Not true.  Sites already track the first node relaying a particular block.  Targeted observation (wiretap) makes the activity even more transparent, when you find a block.

Quote from: jgarzik on April 02, 2013, 11:50:02 PM
Not true.  Sites already track the first node relaying a particular block.  Targeted observation (wiretap) makes the activity even more transparent, when you find a block.

I was referring to the gathering of transactions stage, which is arguably the expensive part, bandwidth wise. If blocks are represented efficiently (eg, list of hashes or deltas against remote expected blocks) then almost all your bandwidth would go on receiving transaction broadcasts, and that doesn't require Tor.

Quote from: Mike Hearn on April 03, 2013, 12:03:11 AM
I was referring to the gathering of transactions stage, which is arguably the expensive part, bandwidth wise. If blocks are represented efficiently (eg, list of hashes or deltas against remote expected blocks) then almost all your bandwidth would go on receiving transaction broadcasts, and that doesn't require Tor.

...which means if I want to prevent mining behind Tor, all I have to do is fill my blocks with transactions that I haven't relayed to them. On the other hand, if it's a rule that you only mine to extend blocks if the transactions were broadcast first, anything that even makes transaction propagation slower, like a bunch of fake nodes, but far from 100%, causes orphan rates to jump. Not to mention I can make it very risky, due to orphaning, to mine blocks containing transactions that a large fraction of the nodes out there have decided to blacklist for whatever reason. You also create new forking risks if transaction propagation is ever prevented, when block propagation isn't.

Personally, I'm fine with a situation where miners anonymously connect to pools that migrate to friendly jurisdictions, but that's just me.  And that's only the worst case scenario if the assumption that Tor can't scale is correct, which I've seen no argument for.  Perhaps retep or jdillon could explain this assumption here?

[tvbcof]

< I apologize, but I must trim a little for brevity. >

I don't believe that things would play out as you imagine as long as the barrier to entry for being a payment processor is low (which, of course, facilitates both competition and the ability for a processor to take bigger risks as they have less to lose.)

The OP has a conception of payment processors being able to demonstrate proof of stake in backing store ownership.  I like this, but also things could work well with 'off chain' meaning being more completely de-coupled or trust-based.

The barrier to become a payment processor that has trust-based relationships with other BTC-banks would be much higher than the barrier to become a Node.

We see in the real world that financial systems built on trust become highly centralized, for reasons independent of regulations, like, well, payment processors, and the way people rely only on the largest networks (Visa/Mastercard/ACH).

This is because trust takes time to build. Centralized BTC-banks are also much more attractive to government as targets for regulatory action, because a takedown of one causes a disruption to the activity it's involved in and is not easily repaired (trust takes time to build).

I do NOT believe that this is why we see centralization in mainstream banking institutions (to the extent that we do.)  The entire mainstream monetary system is built on a system of trusting the power of governments (and having the capitalization to influence them.)  To me, the centralization at the institutional level is most a reflection of the propensity for capital itself to centralize.

A system backed by Bitcoin (and thus, mathematics) will not be prone to the same innate pressures to centralize.  To me, the struggle to keep Bitcoin free of the pernicious effects of monopolization via the necessity for capital intensive systems.

Nodes on the other hand, because they are only one of thousands of nodes storing and propagating the same data, can be taken down without any disruption to the bitcoin economy. A node can also easily pop up to replace any that is taken down. It's only a matter of mustering the technical resources. No one has to be able to trust a node operator for the node to be immediately plugged into the p2p network. Hashes and digital signatures don't lie.

You are basically making my argument for me here.  That is why it is critically important to me that the system remains lite enough such that the barrier to entry is minimal AND that the communication protocol is conduce to reliable function even under coordinated attack by network carriers.  Or at least the kernel of the system (i.e., Bitcoin) because if that remains functional, auxiliary systems can be re-born overnight if they are somehow killed off.

And as the OP points out and is in the process of demonstrating, it is perfectly possible for off-chain providers to provide as much proof (and as little 'trust') as they need in order to attract the users they want.

Lastly, with a very solid core (like Bitcoin of today) the trust between off-chain providers themselves needs to be minimal.  At most they need to trust one another only until the square up, and it would be foolish to put one's solvency at risk by extending this interval.

I seem to have a little more confidence in the potential for crypto-currencies to de-couple themselves from fiat than do you.

I'm not quite sure what you mean here. In which way have I shown little faith in de-coupling from fiat?

That was mostly a mistaken interpretation on my part of what you wrote.

[amincd]

I do NOT believe that this is why we see centralization in mainstream banking institutions (to the extent that we do.)  The entire mainstream monetary system is built on a system of trusting the power of governments (and having the capitalization to influence them.)

A lot of arguments could be made about the causes of centralization, so I'll try to give a more specific example: Visa/Mastercard dominate credit card payments, and would dominate even with no regulations. Networks like this have a network effect. Whoever controls the network then becomes the central power in payment processing.

Merchants have to use their services if they hope to have access to the majority of consumers, and consumers prefer to use their network over those of smaller competitors, even if the smaller competitors offer significant advantages, because whatever those advantages are, they can't make up for the disadvantage of having a smaller network.

It's a snowball effect that's going to lead to a few large networks dominating. Unlike the bitcoin network, a BTC-bank network will require some combination of regulatory approval, trust-building, and fees to the party controlling the network for a participant to join.

This makes it very likely in my opinion to become something similar to the modern payments system, notwithstanding the absence of central bank control over the base currency.

Nodes on the other hand, because they are only one of thousands of nodes storing and propagating the same data, can be taken down without any disruption to the bitcoin economy. A node can also easily pop up to replace any that is taken down. It's only a matter of mustering the technical resources. No one has to be able to trust a node operator for the node to be immediately plugged into the p2p network. Hashes and digital signatures don't lie.

You are basically making my argument for me here.  That is why it is critically important to me that the system remains lite enough such that the barrier to entry is minimal AND that the communication protocol is conduce to reliable function even under coordinated attack by network carriers.  Or at least the kernel of the system (i.e., Bitcoin) because if that remains functional, auxiliary systems can be re-born overnight if they are somehow killed off.

With all due respect, you're missing my point. I'm comparing a few thousand nodes, requiring enterprise-grade hardware to operate, to a large trust-based network of payment processors, that need to stay online and undisturbed for their BTC-credit to maintain its value.

With the latter, there is serious disruption when payment processors within the major network(s) go offline: the BTC-credit they hold can become worthless. And it's not easily repaired (trust-based relationships take time to build).

The former is safe as long as the majority of nodes are operating. If one node goes down, there is no disruption at all to BTC transactions.

Given you need to have one or the other (a network of large nodes or a large trust-based network of payment processors), I would much rather have the former.

And as the OP points out and is in the process of demonstrating, it is perfectly possible for off-chain providers to provide as much proof (and as little 'trust') as they need in order to attract the users they want.

I need MUCH more convincing demonstrations that off chain providers can provide "as much proof" as nodes in the bitcoin network can.

[Peter Todd]

Despite my disagreements with the video I thought it was really well done. I hope somebody does something as equally good for the opposition.

My video was made by stonecanoe - give them a shout.

Personally, I'm fine with a situation where miners anonymously connect to pools that migrate to friendly jurisdictions, but that's just me.  And that's only the worst case scenario if the assumption that Tor can't scale is correct, which I've seen no argument for.  Perhaps retep or jdillon could explain this assumption here?

You know, if you are happy with just assuming pools and nodes in general migrate to friendly jurisdictions you do have a lot of options. But Liberty Reserve just showed how seemingly friendly jurisdictions don't always stay friendly.

Any form of encrypted data transmission, including Tor, has problems in jurisdictions like China which simply throttle encrypted data. The cost to banning encrypted data entirely is just too much, let alone when you include stuff like steganography, (thanks tvbcof) P2P radio links and other exotics. A small blocksize, and off-chain transactions, gives you options.

After all, I've seen people argue that with tx hashes we could easily have a blocksize 5-10x bigger and still mine over anonymous data connections efficiently. That argument is based on taking the average tx size, (160 to 320 bytes) and just passing around tx hashes (32 bytes) instead when you create a block. (this assumes mempools are fairly well synchronized, a problematic assumption, let alone under attack) Well, that's great, but it just gives you 5x-10x better scaling. That doesn't buy you much exponential growth, let alone abusive uses of Bitcoin for things like timestamping that we just can't stop.

What do you do next if 10MB isn't enough? I firmly believe I don't really know what the future of Bitcoin is. I know I don't have a magic crystal ball, but I do believe in having options: http://www.youtube.com/watch?v=2Z902nIOvL0

[d'aniel]

...

Quote from: Mike Hearn on April 03, 2013, 12:03:11 AM
I was referring to the gathering of transactions stage, which is arguably the expensive part, bandwidth wise. If blocks are represented efficiently (eg, list of hashes or deltas against remote expected blocks) then almost all your bandwidth would go on receiving transaction broadcasts, and that doesn't require Tor.

...which means if I want to prevent mining behind Tor, all I have to do is fill my blocks with transactions that I haven't relayed to them. On the other hand, if it's a rule that you only mine to extend blocks if the transactions were broadcast first, anything that even makes transaction propagation slower, like a bunch of fake nodes, but far from 100%, causes orphan rates to jump. Not to mention I can make it very risky, due to orphaning, to mine blocks containing transactions that a large fraction of the nodes out there have decided to blacklist for whatever reason. You also create new forking risks if transaction propagation is ever prevented, when block propagation isn't.

...

I can't see why it matters if an evil miner fills his blocks with transactions that he hasn't relayed.  It doesn't put the Tor miner at any relative disadvantage, as he's still safe receiving the block data without Tor, correct?  Am I missing something here, retep?

[Peter Todd]

I can't see why it matters if an evil miner fills his blocks with transactions that he hasn't relayed.  It doesn't put the Tor miner at any relative disadvantage, as he's still safe receiving the block data without Tor, correct?  Am I missing something here, retep?

That's only true if Tor miners are the majority. If they are the minority, and >50% of hashing power has fast connections, doing so puts that Tor miner at a big disadvantage, and is good for the miners with the fast connection by reducing competition. Keep in mind that it's likely that large well-connected miners will offer SPV node connection services in return for you not giving the competition a chance at mining those transactions and collecting the fees: https://bitcointalk.org/index.php?topic=197169.0

[tvbcof]

I do NOT believe that this is why we see centralization in mainstream banking institutions (to the extent that we do.)  The entire mainstream monetary system is built on a system of trusting the power of governments (and having the capitalization to influence them.)

A lot of arguments could be made about the causes of centralization, so I'll try to give a more specific example: Visa/Mastercard dominate credit card payments, and would dominate even with no regulations. Networks like this have a network effect. Whoever controls the network then becomes the central power in payment processing.

Merchants have to use their services if they hope to have access to the majority of consumers, and consumers prefer to use their network over those of smaller competitors, even if the smaller competitors offer significant advantages, because whatever those advantages are, they can't make up for the disadvantage of having a smaller network.

It's a snowball effect that's going to lead to a few large networks dominating. Unlike the bitcoin network, a BTC-bank network will require some combination of regulatory approval, trust-building, and fees to the party controlling the network for a participant to join.

This makes it very likely in my opinion to become something similar to the modern payments system, notwithstanding the absence of central bank control over the base currency.

Natural monopolies are...well...natural in some environment.  Namely one's where the barrier to entry is high.  They are also artificially created when possible.  Visa/MC are probably a little bit of both.

I conjecture that efforts based on a freely accessible Bitcoin network will have a very low barrier to entry and can arrange an endless array of unique incentives so I very much doubt that they would fall victim to the natural monopoly effect.

We may have to agree to dis-agree on this, but it is pretty clear to me that we are both handwaving to some extent.

With all due respect, you're missing my point. I'm comparing a few thousand nodes, requiring enterprise-grade hardware to operate, to a large trust-based network of payment processors, that need to stay online and undisturbed for their BTC-credit to maintain its value.

With the latter, there is serious disruption when payment processors within the major network(s) go offline: the BTC-credit they hold can become worthless. And it's not easily repaired (trust-based relationships take time to build).

The former is safe as long as the majority of nodes are operating. If one node goes down, there is no disruption at all to BTC transactions.

Given you need to have one or the other (a network of large nodes or a large trust-based network of payment processors), I would much rather have the former.

OK, even if Bitcoin growth happened to stop at a point when a few thousand entities could still operate it...

As I've mentioned on previous threads, there is much more value in having a broad footprint in the network and using it to harvest business intelligence information than there would be in collecting penny-ante fees.  So, entities who have a realistic ability to tap into this source of value will be able to subsidize the services they offer.  This will drive out competition and create a vicious circle of centralization. 

For an example of this, consider how e-mail has evolved over the past 10 years ago.  Relatedly, look at how much more centralized the e-mail solutions most of us use have become.  You think that this happens just because some CEO felt like giving something back to the community?  Think again.

And as the OP points out and is in the process of demonstrating, it is perfectly possible for off-chain providers to provide as much proof (and as little 'trust') as they need in order to attract the users they want.

I need MUCH more convincing demonstrations that off chain providers can provide "as much proof" as nodes in the bitcoin network can.

I'm looking forward to his work along these lines.