Pages:
Author

Topic: New video: Why the blocksize limit keeps Bitcoin free and decentralized (Read 15201 times)

legendary
Activity: 1260
Merit: 1002
Very nice video, helped me grasp the matter quite fairly.
Funny that it was done prior Gavin's current testing and pushing such fork.
hero member
Activity: 490
Merit: 500
I'm inclined to agree.  I like hosting a full node on my computer, but I've only had it for a couple months and I'm already seeing exponential growth in the size of the blockchain.  If the growth of the blockchain continues at the current rate, it'll be too big to host on most people's computers in a few years.

The blockchain will never be too big to store on desktop computers.  Moore's law applies to hard drives.  3tb drives are now under $100, and in two years, it will be 6tb.  The blockchain will only continue an exponential path while bitcoin is adding new users.  Once we reach target market size or a blocksize cap, the blockchain will increase in size linearly.  In 20 years, people will be able to store the blockchain on their phone and download the whole thing in a few hours.

Ultimately, I think there are two different issues with the blockchain.  I agree with you about Moore's law - I have a 250 GB hard drive and I'm not too concerned about the blockchain outgrowing my hard drive during the time while I'm still using this computer.  Although I'd be curious to see where you found an under $100 price for 3 TB hard drives, I looked on Amazon recently for hard drives in the 3 -4 TB range, and they were all $150 - $180.

The bigger issue isn't storing the blockchain but the initial download & verification process.  It took me three days to do this back when it was a 7 GB blockchain, it was in the 9 GB range last I checked.  People usually have much smaller bandwidth than they do hard drive space.
sr. member
Activity: 328
Merit: 250
I'm inclined to agree.  I like hosting a full node on my computer, but I've only had it for a couple months and I'm already seeing exponential growth in the size of the blockchain.  If the growth of the blockchain continues at the current rate, it'll be too big to host on most people's computers in a few years.

The blockchain will never be too big to store on desktop computers.  Moore's law applies to hard drives.  3tb drives are now under $100, and in two years, it will be 6tb.  The blockchain will only continue an exponential path while bitcoin is adding new users.  Once we reach target market size or a blocksize cap, the blockchain will increase in size linearly.  In 20 years, people will be able to store the blockchain on their phone and download the whole thing in a few hours.
hero member
Activity: 490
Merit: 500
I'm inclined to agree.  I like hosting a full node on my computer, but I've only had it for a couple months and I'm already seeing exponential growth in the size of the blockchain.  If the growth of the blockchain continues at the current rate, it'll be too big to host on most people's computers in a few years.
full member
Activity: 182
Merit: 100
legendary
Activity: 4592
Merit: 1276

I anticipate that what will develop will be a selection of turn-key engines which will gaurentee to people like me that people like you are running a service which is highly likely to be legitimate and will achieve the result I want.

Are you kidding me? You're anticipating turn-key engines that solve internet security once and for all and allow anyone to build a BTC-accepting site with minimal cost like that and have everyone trust it?

'satoshi' did it.  It's much easier when standing on the shoulders of giants.

You also evaded my point about BTC-credit not being fungible:

Ya.  I explained/apologized in advance.
hero member
Activity: 772
Merit: 501
Quote
If I, as a plain-jane user like someone's micro-blog and wishes to give a micro-transaction in appreciation, I would be happy to see a known entity like 'bittip' or one which I could easily verify as an entity who uses best-practice technology which precludes fraud/theft and produces some reasonable level of transparency.  Happy enough to vastly increase my deployment of 'tips'.

That's tangential and not an explanation for how I would run my site as it is now.

My site doesn't have professional security measures/best-practices in places. It's an experimental site I run on the side so users do not trust it with more than $20 worth of BTC deposits per week cumalatively. With a 1 MB block size limit and $20 transaction fees, my site would not be possible.


I anticipate that what will develop will be a selection of turn-key engines which will gaurentee to people like me that people like you are running a service which is highly likely to be legitimate and will achieve the result I want.

Are you kidding me? You're anticipating turn-key engines that solve internet security once and for all and allow anyone to build a BTC-accepting site with minimal cost like that and have everyone trust it?

You also evaded my point about BTC-credit not being fungible:

To explain further, let's say one of my users has BTC deposited with MtGox. He wants to deposit some BTC in my site, but doesn't want to pay a $20 transaction fee, so provides a "reference" to some of the BTC he has stored at MtGox.

I therefore receive some of the BTC-credit (promissory-notes), as opposed to actual BTC, that is backed by the full faith and credit of MtGox.

Overtime, the BTC-credit deposited at btctip would become an assortment of promissory notes from various BTC-banks, each with a different value depending on how much the backing bank is trusted.

If one of my users 'tips' another user, my site would have to try to determine which type of BTC-credit to transfer. Remember: I don't have any actual BTC deposited at my site, just promissory notes, issued by different BTC-banks. There would be no obvious way to do this. My site would not be possible.


My friend, your solution would not work. We cannot have a 1 MB block size limit and $20 transaction fees and see a successful Bitcoin. If you can't use Bitcoin transactions, then it's not Bitcoin. All of the innovation and flexibility that one sees with BTC e-wallets springs from the low cost to deposit and withdraw BTC to/from those e-wallets. Take that away, and you have the modern financial system and all of its inefficiencies and limitations.
legendary
Activity: 4592
Merit: 1276
...
If we're thinking about system-level attacks, the relevant people to think about are the people who actually will be doing the mining,
...

We are probably not.

The attacks which concern me the most have to do with deep packet inspection, near real-time network mesh analysis, and targeted packet filtering.  And relatedly, discrimination of encrypted traffic that was not encrypted in an 'authorized' manner.

I don't believe that "Moore's Law" is really very applicable to Bitcoin since Bitcoin is very much in it's infancy (among other reasons) but I won't overload this note with that topic other than to say that "Moore's Law" almost certainly helps the likely attackers of Bitcoin more than it helps the defenders.

sr. member
Activity: 352
Merit: 250
https://www.realitykeys.com
Now we have exponentially growing bandwidth capacity that WE CAN'T USE. nice! Tell me, how many locales in the americas can you get 1 Gb/s fiber?

I'm in the rural Western US.  The only hard-line I have is POTS which on a good day gets 19,200 kbps.

Since mid-2012 I've been able to have a surprisingly usable satellite connection (Exede) but it goes down for long-ish periods quite regularly and for $80/mo I am capped at 15GB per month.  It also has typical satellite latency which impact certain software architectures.

---

I'm not suggesting that Bitcoin be designed around allowing me to be a peer (though it would be nice if I could be.)  The main reasons for this are that it would be to limiting to be realistic, and it would not offer the protection against system level attack to make it worthwhile.

Right, I wouldn't be surprised if Grue's ADSL connection was hurting the ability to mine profitably even with the current block sizes. If you have to pay the same as your competitors for hardware and electricity, a competitive market should punish even a fairly small persistent disadvantage that your urban competitors don't have. In the same way, there will be large parts of the world (including here in Tokyo) where you can't mine profitably long-term in a seriously competitive market because electricity costs too much.

If we're thinking about system-level attacks, the relevant people to think about are the people who actually will be doing the mining, and we should assume that in a competitive market, everything about their setup will be reasonably well optimized. That makes high-end connections the relevant factor here, and they may even be growing _faster_ than Moore' Law.
http://www.sqw.co.uk/sqw-commentary/the-urban-rural-digital-dividewider-and-wider

(Overall bandwidth seems to be growing a bit slower than Moore's Law, presumably because the connections of unlucky people like tvbcof and grue in rural areas of countries with disfunctional telecom markets pull down the average...)
legendary
Activity: 4592
Merit: 1276
Now we have exponentially growing bandwidth capacity that WE CAN'T USE. nice! Tell me, how many locales in the americas can you get 1 Gb/s fiber?

I'm in the rural Western US.  The only hard-line I have is POTS which on a good day gets 19,200 bps.

Since mid-2012 I've been able to have a surprisingly usable satellite connection (Exede) but it goes down for long-ish periods quite regularly and for $80/mo I am capped at 15GB per month.  It also has typical satellite latency which impact certain software architectures.

---

I'm not suggesting that Bitcoin be designed around allowing me to be a peer (though it would be nice if I could be.)  The main reasons for this are that it would be to limiting to be realistic, and it would not offer the protection against system level attack to make it worthwhile.

  edit: bps, not kbps.
legendary
Activity: 4592
Merit: 1276
Quote
If I, as a plain-jane user like someone's micro-blog and wishes to give a micro-transaction in appreciation, I would be happy to see a known entity like 'bittip' or one which I could easily verify as an entity who uses best-practice technology which precludes fraud/theft and produces some reasonable level of transparency.  Happy enough to vastly increase my deployment of 'tips'.

That's tangential and not an explanation for how I would run my site as it is now.

My site doesn't have professional security measures/best-practices in places. It's an experimental site I run on the side so users do not trust it with more than $20 worth of BTC deposits per week cumalatively. With a 1 MB block size limit and $20 transaction fees, my site would not be possible.


I anticipate that what will develop will be a selection of turn-key engines which will gaurentee to people like me that people like you are running a service which is highly likely to be legitimate and will achieve the result I want.  As a bonus, it will be free to you.  You can then focus on value-adds which are important to your customers.

Everyone wants a free ride on the blockchain.  If it's offered I'll hop right on (as outlined in my sig.)

At this point I still hold hope that Bitcoin proper can evolve to form the backbone of a solution that we, as citizens of Planet Earth, badly need.  If Bitcoin can move to that role without being 'PayPal-ified' by well capitalized corporate interests it will be better for everyone.  Or everyone I care about, at least, and that certainly includes efforts such as 'bittip'.

---

I'm sorry to snip the remaining, but I'm exasperated with this conversation and I don't believe that it is achieving much for anyone at this point.  Once some other material which explains the basic problems, principles, definitions, and possibilities makes it's way around, we can re-visit some of this stuff.  These were only touched on in the vid due to lack of time and lack of development.

sr. member
Activity: 328
Merit: 250
People have been calling for an end to moore's law for decades and been wrong.  Besides, the only important parts of moore's law here is hard drive capacity and internet bandwidth, not CPUs here.  These are not going to slow down their exponential growth any time in the next decades.  Hard drive capacity in particular seems likely to have radical breakthroughs that smash the pace of moore's law.
Only problem: moore's law doesn't apply to bandwidth capacity. Case in point: I'm still stuck with my ADSL 1 internet. My other option is VDSL or DOCIS 3.0 but they are only offered by incumbents and they have a cap of 80 GB/month.

Don't be ridiculous.  Moore's law applies to bandwidth just fine.  In fact it seems to be speeding up.  The law does not fail just because when fiber is rolled out, it is not installed in your particular house.
http://www.cabletechtalk.com/broadband-internet/broadband/broadband-speed-and-moores-law-a-response-to-robb-topolski/

1980:  300 baud
1990: 14.4k
2000: 384k dsl
2013: 1000mb fiber
Now we have exponentially growing bandwidth capacity that WE CAN'T USE. nice! Tell me, how many locales in the americas can you get 1 Gb/s fiber?
Are you ******* serious dude?  You obviously don't have a clue what moore's law is.  Hardly anyone had 384k DSL at the turn of the century either.  Change it to 2013: 20mb.  It still follows moore's law.  Did you even see the freaking graph I posted?  World internet bandwidth has gone from less than 1,000 Gb/s in 2001 to 80,000 Gb/s in 2011.  It's too bad they missed your house, but stfu about it.
hero member
Activity: 772
Merit: 501
You're not addressing my points here:

...
My website btctip.com is a good example of an amateur startup that couldn't operate in this high-powered expensive BTC world. On a very active week, I have maybe $20 worth of transactions into/out-of my site. Without low-cost network transactions, these tiny deposits and withdrawals would not be possible.
...

Good heavens!  You are already an off-chain processor.  This is great!  It's pretty much sinking in to most people by this time that micro-payments are unrealistic, and a lot of us (including myself) are disappointed...

But I still need low cost transactions to allow users to deposit and withdraw, otherwise even the <$20 worth of deposit/withdrawal activity that happens per week wouldn't happen.

Quote
If I, as a plain-jane user like someone's micro-blog and wishes to give a micro-transaction in appreciation, I would be happy to see a known entity like 'bittip' or one which I could easily verify as an entity who uses best-practice technology which precludes fraud/theft and produces some reasonable level of transparency.  Happy enough to vastly increase my deployment of 'tips'.

That's tangential and not an explanation for how I would run my site as it is now.

My site doesn't have professional security measures/best-practices in places. It's an experimental site I run on the side so users do not trust it with more than $20 worth of BTC deposits per week cumalatively. With a 1 MB block size limit and $20 transaction fees, my site would not be possible.

Quote
Bitcoin seems like some magic bullet which makes everything free and easy at this point, but that is mainly an artifact of it's being a tiny tiny spec in the global economy at this time.  It will become more expensive and complex as it grows.

Transactions don't need to become more expensive as it grows. We can simply not have a 1 MB block size limit.

Quote
We have entities chomping at the bit to subsidize the cost of this growth because they would like to capitalize on some of the value streams it offers.

There's no evidence this is happening.

Quote
In practice, there would probably be three options from the perspective of the micro-blogger.  (Using $ terms for simplicity.)

 - If the tips are, in the range of $0->$10 per month, just provide reference(s) to other processor(s) which are used.

First of all, as I've mentioned before, it's not "other processor(s)" we're talking about, it's "other banks". If btctip were to provide a reference to a BTC-bank, it would cease to be dealing in BTC. Instead, it would be dealing with BTC-credit.

To explain further, let's say one of my users has BTC deposited with MtGox. He wants to deposit some BTC in my site, but doesn't want to pay a $20 transaction fee, so provides a "reference" to some of the BTC he has stored at MtGox.

I therefore receive some of the BTC-credit (promissory-notes), as opposed to actual BTC, that is backed by the full faith and credit of MtGox.

Overtime, the BTC-credit deposited at btctip would become an assortment of promissory notes from various BTC-banks, each with a different value depending on how much the backing bank is trusted.

If one of my users 'tips' another user, my site would have to try to determine which type of BTC-credit to transfer. Remember: I don't have any actual BTC deposited at my site, just promissory notes, issued by different BTC-banks. There would be no obvious way to do this. My site would not be possible.

In this economy, only the largest BTC-banks would have universally trusted BTC promissory notes, with most users opting to their credit. This would create a tendency to more centralization of the BTC-economy around a few large financial institutions, with small players like my website being completely dependent on them.

The power of the largest BTC-banks would be like the power large financial institutions have today. They would have the power to exclude anyone they wanted from using their services, and those excluded would be at a severe disadvantage relative to those approved by the large BTC financial institutions.

This is why I see a permanent 1 MB block size, that forces transaction fees to $20, as going against everything that makes BTC useful.
legendary
Activity: 2058
Merit: 1431
People have been calling for an end to moore's law for decades and been wrong.  Besides, the only important parts of moore's law here is hard drive capacity and internet bandwidth, not CPUs here.  These are not going to slow down their exponential growth any time in the next decades.  Hard drive capacity in particular seems likely to have radical breakthroughs that smash the pace of moore's law.
Only problem: moore's law doesn't apply to bandwidth capacity. Case in point: I'm still stuck with my ADSL 1 internet. My other option is VDSL or DOCIS 3.0 but they are only offered by incumbents and they have a cap of 80 GB/month.

Don't be ridiculous.  Moore's law applies to bandwidth just fine.  In fact it seems to be speeding up.  The law does not fail just because when fiber is rolled out, it is not installed in your particular house.
http://www.cabletechtalk.com/broadband-internet/broadband/broadband-speed-and-moores-law-a-response-to-robb-topolski/

1980:  300 baud
1990: 14.4k
2000: 384k dsl
2013: 1000mb fiber
Now we have exponentially growing bandwidth capacity that WE CAN'T USE. nice! Tell me, how many locales in the americas can you get 1 Gb/s fiber?
sr. member
Activity: 352
Merit: 250
https://www.realitykeys.com
People have been calling for an end to moore's law for decades and been wrong.  Besides, the only important parts of moore's law here is hard drive capacity and internet bandwidth, not CPUs here.  These are not going to slow down their exponential growth any time in the next decades.  Hard drive capacity in particular seems likely to have radical breakthroughs that smash the pace of moore's law.
Only problem: moore's law doesn't apply to bandwidth capacity. Case in point: I'm still stuck with my ADSL 1 internet. My other option is VDSL or DOCIS 3.0 but they are only offered by incumbents and they have a cap of 80 GB/month.

Don't be ridiculous.  Moore's law applies to bandwidth just fine.  In fact it seems to be speeding up.  The law does not fail just because when fiber is rolled out, it is not installed in your particular house.
http://www.cabletechtalk.com/broadband-internet/broadband/broadband-speed-and-moores-law-a-response-to-robb-topolski/

1980:  300 baud
1990: 14.4k
2000: 384k dsl
2013: 1000mb fiber

Nielsen reckons network growth is a bit slower, but still exponential. He has network speed increasing at 50% per year, while Moore's Law works out at about 60% per year.
http://www.nngroup.com/articles/law-of-bandwidth/
sr. member
Activity: 328
Merit: 250
People have been calling for an end to moore's law for decades and been wrong.  Besides, the only important parts of moore's law here is hard drive capacity and internet bandwidth, not CPUs here.  These are not going to slow down their exponential growth any time in the next decades.  Hard drive capacity in particular seems likely to have radical breakthroughs that smash the pace of moore's law.
Only problem: moore's law doesn't apply to bandwidth capacity. Case in point: I'm still stuck with my ADSL 1 internet. My other option is VDSL or DOCIS 3.0 but they are only offered by incumbents and they have a cap of 80 GB/month.

Don't be ridiculous.  Moore's law applies to bandwidth just fine.  In fact it seems to be speeding up.  The law does not fail just because when fiber is rolled out, it is not installed in your particular house.
http://www.cabletechtalk.com/broadband-internet/broadband/broadband-speed-and-moores-law-a-response-to-robb-topolski/

1980:  300 baud
1990: 14.4k
2000: 384k dsl
2013: 1000mb fiber
sr. member
Activity: 352
Merit: 250
https://www.realitykeys.com
We should only expand block size when there's a legitimate bottleneck. Satoshidice and other "on the chain" services does not count. Neither do small value transactions that can be settled by off the chain transaction providers. If you take away all of those, we have plenty of room to grow.

Out of interest, what's the specific cost per transaction, in USD, that you think will price out Satoshi Dice while keeping things that are "legitimately" on the block-chain viable?
legendary
Activity: 2058
Merit: 1431
We should only expand block size when there's a legitimate bottleneck. Satoshidice and other "on the chain" services does not count. Neither do small value transactions that can be settled by off the chain transaction providers. If you take away all of those, we have plenty of room to grow.

People have been calling for an end to moore's law for decades and been wrong.  Besides, the only important parts of moore's law here is hard drive capacity and internet bandwidth, not CPUs here.  These are not going to slow down their exponential growth any time in the next decades.  Hard drive capacity in particular seems likely to have radical breakthroughs that smash the pace of moore's law.
Only problem: moore's law doesn't apply to bandwidth capacity. Case in point: I'm still stuck with my ADSL 1 internet. My other option is VDSL or DOCIS 3.0 but they are only offered by incumbents and they have a cap of 80 GB/month.
sr. member
Activity: 328
Merit: 250
People have been calling for an end to moore's law for decades and been wrong.  Besides, the only important parts of moore's law here is hard drive capacity and internet bandwidth, not CPUs.  These are not going to slow down their exponential growth any time in the next decades.  Hard drive capacity in particular seems likely to have radical breakthroughs that smash the pace of moore's law.

Flying cars seems like an ironic example to choose.  We already have flying cars.  They will get popular once some infrastructure is built and the price comes down a bit.
http://www.youtube.com/watch?v=nnF2yua4KIw
http://www.youtube.com/watch?v=CajAq6ndJYE
http://www.youtube.com/watch?v=FY85eExk7Zo

Also, by 2030 we will be able to fly London>Sydney in 1 or 2 hours with hypersonic aircraft.

Why do you think censorship-proof network bandwidth doesn't scale with technology improvements?  You think TOR would be just as fast if we were all running 14.4k modems?  You think TOR will be the same speed when we all have 1gb internet connections?
legendary
Activity: 1120
Merit: 1150
There was a recent article in Bitcoin Magazine that analysed the blockchain size, but most of their articles are not available online.  The article looked at some worst case scenarios, and even using really conservative estimates, 20 years from now people will easily be able to store the whole blockchain on their phone and download the whole thing in a few hours.

Bitcoin Magazine is assuming technological growth continues to grow exponentially. Unfortunately that's an extremely optimistic assumption at best. For instance, the width between the smallest features on transistors will be on the order of 10-20 atoms in just a few more years, and frankly no-one in the industry has any clue how we're going to do much better than that, or even if we can. (I work as an electronics designer) Additionally the cost of chip fabs is already to the point where the whole world economy can only support about 10 top-of-the-line fabs, and that number has historically gotten smaller for every technological iteration.

You remind me of the people who were predicting flying cars back in the 60's on the basis that airplane technology had been improving exponentially prior to the 60's. Everything looked great... until they hit the physical limits of the technology.

In any case, focusing on raw technology is missing the point: what matters is censorship-proof network bandwidth. Unfortunately that doesn't scale with technology improvements.
Pages:
Jump to: