I wouldn't mind having $5 of my money held at Apple/Google/VISA/Mastercard/BitPay (and I wouldn't be sad of losing $5 if any of these companies go bankrupt).

Actually I had in mind creating a centralized version of Bitcoin for ultra-fast payments. With keeping all addresses on SSDs, asking for 1 cent / address month, 1 cent / transaction should be possible to reach even with 6x replication. Companies could compete in price as long as the API is standardized. Automatic top-up should be simple as well.

On Wed, Apr 23, 2014 at 10:53 PM, Gregory Maxwell <gmaxwell@gmail.com> wrote:

On Wed, Apr 23, 2014 at 1:44 PM, Adam Ritter <aritter@gmail.com> wrote:
> Isn't a faster blockchain for transactions (maybe as a sidechain) solving
> the problem? If there would be a safe way for 0-confirmation transactions,
> the Bitcoin blockchain wouldn't even be needed.

Large scale consensus can't generally provide instantly irreversible
transactions directly: Increasing the block speed can't help past the
point where the time starts getting close to the network diameter...
you simply can't tell what a consensus of a group of nodes is until
several times the light cone that includes all of them. And if you
start getting close to the limit you dilute the power working on the
consensus and potentially make life easier for a large attacker.

Maybe other chains with different parameters could achieve a different
tradeoff which was better suited to low value retail transactions
(e.g. where you want a soft confirmation fast). A choice of tradeoffs
could be very useful, and maybe you can practically get close enough
(e.g. would knowing you lost a zero-conf double spend within 30
seconds 90% of the time be good enough?)... but I'm not aware of any
silver bullet there which gives you something identical to what a
centralized service can give you without invoking at least a little
bit of centralization.