<infinisil>
I've had my music library in beets for like 2 years now, and I still haven't rated all songs
<infinisil>
Recently I'm picking up again, now I only have ~350 left
<infinisil>
What I learned from this: Even if you think a song is good, that doesn't mean you have to like it
<infinisil>
So often I go "Oh that's a decent song", but I don't like it for whatever reason. I rate these songs low because of this
<infinisil>
Also: Don't be afraid to use the full spectrum of ratings (for me that's 1-10)
<gchristensen>
anyone seen a recent FF have horrible video playback framerates?
<samueldr>
yes
<gchristensen>
oh good
<samueldr>
generally bad framerates, not only videos
<samueldr>
gifs in page make scrolling chuggy as it can be
<gchristensen>
if I pop out the video player, it works much better
<gchristensen>
samueldr: setting gfx.direct3d11.use-double-buffering to false - does this improve it?
<samueldr>
that sounds windowsey
<samueldr>
I don't think so, but I don't have a good testing page to test conclusively
<gchristensen>
me either
<evanjs>
I have the same problem with ratings but I'm so much worse at it. This can be seen in my Apple Music/iTunes library and MAL watch/reading lists
<evanjs>
One of the reasons I appreciate the focus/move to like/dislike based ratings. I'm much more likely to, at the very least, answer that
<infinisil>
evanjs: Yeah, I've been thinking about how to have better rating systems for a while
<infinisil>
E.g. one that's based on "which of these two songs do you like more", from which the system can generate a ranking over time
<gchristensen>
how do you handle uncomparable songs
<infinisil>
gchristensen: Hm I guess there needs to be a neutral option
<gchristensen>
more, the songs can't be strictly ordered
<infinisil>
Ah well it doesn't have to be that, an approximate ordering will be good enough
<infinisil>
Or should
<gchristensen>
which do I like better: Her Majesty, The Beatles ...or... She Don't Use Jelly, The Flaming Lips -- neither, they're not even close to the same category
<infinisil>
Haven't thought much about categorization
evanjs has joined #nixos-chat
<infinisil>
gchristensen: Ah! You just answer "uncomparable", then the system knows that these songs are in different categories
<gchristensen>
:)
<infinisil>
And over time it collects data on which categories there are (with some vector space probably)
<gchristensen>
a nice idea
<infinisil>
This comparison-based rating system makes for an interesting problem then: What song does the system choose to play next?
<gchristensen>
especially since there is almost certainly a route of comparable songs from Her Majesty to She Don't Use Jelly
<infinisil>
I don't know either, but yeah!
<jackdk>
I'm looking forward to this system finding a cycle of preferences and looping forever
<infinisil>
jackdk: An important part would be to have answers be probabilistic to prevent this
<infinisil>
So it could tweak the answers a bit to make it work
* gchristensen
idly wonders where in the vector space is Once in a Lifetime
<infinisil>
And also important: Preferences change over time and moods, ideally it can detect your mood based on your answers and e.g. play things you liked during that mood in the past
<infinisil>
I think it's important that something like this should include that "you are using other people's free time" (it does include that, but only as a side remark)
<infinisil>
Okay I just read through http://slash7.com/2006/12/22/vampires/. zeta is a Help Vampire, no doubt, and that article is the best for them to read
<infinisil>
It's amazing how well that article fits
<jackdk>
infinisil: I agree with your assessment of .../getting_answers.html . The author has an email link at the bottom for feedback; have you sent any?
<infinisil>
Nope
<infinisil>
I'm not sure if this fits as a point in that article
<infinisil>
The slash7 article is perfect for this case though
srhb has quit [Read error: Connection reset by peer]
zimbatm has quit [Read error: Connection reset by peer]
srhb has joined #nixos-chat
peel has joined #nixos-chat
zimbatm has joined #nixos-chat
elvishjerricco has quit [Ping timeout: 240 seconds]
elvishjerricco has joined #nixos-chat
Synthetica has joined #nixos-chat
waleee-cl has joined #nixos-chat
endformationage has joined #nixos-chat
tmplt has joined #nixos-chat
waleee-cl has quit [Quit: Connection closed for inactivity]
<__monty__>
clever: Do you have any insights into the IOHK CIs for haskell.nix?
<clever>
__monty__: in what area of it?
<__monty__>
I'd like to pin haskell.nix to leverage the cachix cache. So I figure I'll take the most recent succesful CI build. But there's 4 listed on github, hercules derivation, hercules evaluation, buildkite and iohk-hydra. I don't know which one I should look at.
<__monty__>
I assume hercules because cachix but the latest passing build there is pretty old.
<__monty__>
And I can't access the buildkite logs because I don't have a login.
<clever>
__monty__: the hydra based stuff is all served via hydra's cache
<clever>
i think hercules goes into cachix
<Taneb>
Would you say "withe is a typo of with the" or "withe is a typo for with the"?
<__monty__>
I'm not sure what the iohk-hydra builds actually.
<clever>
__monty__: i think stack2nix and haskell.nix are going to be hard to cache for under any conditions, because you can change core haskell libs
<clever>
which causes a rebuild of the entire haskell ecosystem
<clever>
the most you can do, is fire up a hydra pointed to your specific project, that will obey your stack.yaml
<__monty__>
clever: I don't think I need to do that currently though, and I'd *much* rather use cached artifacts.
<clever>
__monty__: ive not looked at the haskell.nix ci itself much, ive mostly been focusing on the down-stream stuff like cardano-node
<__monty__>
I suspect no one on irc has >.<
<clever>
__monty__: there is #haskell.nix
<clever>
ah, your already there
<__monty__>
Yeah, I've asked there yesterday and today : )
<__monty__>
I remembered you're an IOHK ops so I figured I'd try to snipe ; )
<clever>
__monty__: angerman is the main guy in the haskell.nix area
tmplt has quit [Quit: leaving]
drakonis has joined #nixos-chat
mudri has quit [Read error: Connection reset by peer]
mudri has joined #nixos-chat
nh2 has quit [Read error: Connection reset by peer]
nh2 has joined #nixos-chat
<Church->
Okay got my pine phone
<Church->
And back in the box it goes for a few weeks
wildtrees has joined #nixos-chat
<eyJhb>
samueldr: Have you looked at COSMO COMMUNICATOR ?
<samueldr>
I loaded up images on the information super highway and my peepers saw those images
<samueldr>
so, "yes"
<samueldr>
but I figure that's not what you want to know
<eyJhb>
Just regarding NixOS for mobile, seems pretty cool
<eyJhb>
And I am somewhat confused by the first sentence :p
<samueldr>
"saw pictures online with my eyes"
<eyJhb>
Ahh
<eyJhb>
But it seems super cool if it could be NixOSIFIED
<samueldr>
I don't have the device, though it's likely possible to do because of how unclosed it is
<samueldr>
(not used open as it's not necessarily open)
<eyJhb>
A PhD student at my university just got his, but he is not very NixOS friendly :/ Lets see if we can get you the device ;)
<gchristensen>
it has been a long time since a programming thing made me want to cry
<__monty__>
Tell us about this gkryptonite.
<gchristensen>
Rust's Futures
<__monty__>
Haven't had the pleasure.
<__monty__>
More vexing than the borrowing system?
<gchristensen>
I guess so
<gchristensen>
at least, backwhen I was learning Rust errors with the borrow checker had errors I could reason through
<gchristensen>
but Rust doesn't have good Future errors yet
<ar>
eyJhb: i use it as my main phone, and i like the keyboard (hardware wise), i hate the tiny pointless differences between us/pl hwkeyboard layouts (you can switch between keyboard layouts easily)
<ar>
eyJhb: some android apps are troublesome as they force everything into landscape
<ar>
eyJhb: so when an app has a long list of items somewhere where the developers didn't expect you needing to scroll…
<ar>
eyJhb: and i'm salty about how bad planetcom is with regards to providing software updates for their devices
<ar>
(the reason for that may be that they've chosen to use mediakek SoCs, but i don't think anyone's forcing them to use mediakek)
<ar>
eyJhb: the external screen is nice for finally being able to tell who's calling you (as opposed to what you had with geminipda), but it would be awesome to be able to turn off touchscreen on it
<ar>
eyJhb: because as it is i end up triggering stuff while the phone is in my pockets, like pausing music or skipping tracks
<ar>
eyJhb: also, as the external screen is handled through a secondary microcontroller (some stm32f103, afaik), instead of the main cpu, it is a bit laggy
<__monty__>
Does "external" screen mean screen on the back?
<ar>
yup
<eyJhb>
ar: and front?
<__monty__>
Wait, how is a screen on the back handier to see who's calling?
<samueldr>
the device is a foldable
<samueldr>
like a laptop
<__monty__>
Wait, we're not talking about a pine phone?
<ar>
no
<joepie91>
cosmo communicator, from the sounds of it
<ar>
yup
<__monty__>
Oh. Assumed cosmo was a version of android.
<samueldr>
it's an android phone
<ar>
eyJhb: the front screen (or, rather, "internal", if that makes sense) is connected to the main cpu as you'd usually expect on a phone
<ar>
eyJhb: except instead of doing a tablet build (of android; what i'm describing here is - iirc - build-time configurable) and letting the user change between landscape and portrait through "notification bar settings" (let's face it, most of the apps that are buggy on a device like that aren't going to get fixed), they've opted for doing a phone build reporting the screen as having a 2160x1080
<ar>
resolution (as opposed to 1080x2160 it'd be on a "normal phone") and absolutely no way to let you change orientation
<ar>
depending on the apps you use, this may or may not be a big problem; the things i use most - weechat-android, termux, signal - behave well, but the ones i use only occasionally - like an app for planning routes w/ public transit, or a taxi hailing app - are a bit painful
<ar>
and at the price i got it from the indigogo campaign - $500 or $550 - it's a nice device, but at the prices it's currently at ($750), and because it's their second device in the same formfactor and sharing most of the issues between them, i'd expect at least some of the issues here to, well, not be here
<eyJhb>
ar: yeah okay, I can see how all that is very much con, but still work-in-progress,but sucks if they haven't fixed previous flaws
<eyJhb>
I once had the Asus Padfone 5 (I think), riddled with flaws....
<ar>
well, they've fixed two biggest issues with the previous one (which i also have) - you can see who's calling you before opening the phone, and it has a fingerprint reader now
Peetz0r has quit [Quit: de aardlek test je met je tong, toch?]
Peetz0r has joined #nixos-chat
<infinisil>
Argh why does github have to be a shitty SPA
<infinisil>
Or is it?
<infinisil>
I think it is
<infinisil>
All these glitches where the next page doesn't load correctly are annoying
<infinisil>
Or how it displays a progressing loading bar at the top, even though I don't even have an internet connection
<samueldr>
it's not really an SPA, it's "turbolinks"
<samueldr>
or whatever new incarnation
<samueldr>
and their implementation is bad™
<samueldr>
my main gripe, other than being stuck in an infinit load, is how often it drops navigations from the history
<infinisil>
Ugh
<__monty__>
I love how my history always only remembers specific files deep in repos, rather than the root. All those extra clicks are glorious...
<samueldr>
not even thinking about the "browser history" but "back navigation history" here
<samueldr>
like using "back" goes two pages back instead of one
<__monty__>
That's part of browser history, isn't it?
<__monty__>
I don't usually have trouble with that but maybe I've unconsciously taught myself to just click to where I believe back should take me?
<__monty__>
nn, peoples
<samueldr>
yeah, part of the history, but pointing out the problem
<samueldr>
and no, not part of "history" the html5 api
__monty__ has quit [Quit: leaving]
<joepie91>
infinisil: that fake progress bar is a shitty workaround for breaking the browser's UI and not having access to the needed APIs to determine progress reliably
<joepie91>
basically all the SPA-y progress bars lie to you
<joepie91>
they generally use animation to make it look like they're real progress bars
<joepie91>
starting out fast and then gradually slowing down
<infinisil>
So annoying..
<infinisil>
Makes me think I have internet connection, only to realize it never gets to the end..
<samueldr>
joepie91: amen
<samueldr>
if there was an API that gave us a percent amount, or bytes amount (with proper cross-domain auth) that would at least help with that... but it's still a crap patch on top of a feature that was mostly removed from browsers
<samueldr>
(there's no status bar, and load progress anymore in mainstream browsers)
<samueldr>
(and anyway with all that async ajax stuff it wouldn't mean as much)
<joepie91>
there's still a loading indicator though
<samueldr>
yeah
<samueldr>
it would be so much better if all network event triggered it
<joepie91>
samueldr: that used to be the case and then people complained that they couldn't do longpolling-esque things without the page perpetually looking like it's loading
<samueldr>
but... it is!
<samueldr>
(and yeah, I know about it all)
<infinisil>
Would get some push-back against shitty websites!
<samueldr>
browsers a re *user agents*, not *website agents*
Tjowers has joined #nixos-chat
Yaniel has joined #nixos-chat
<Tjowers>
Ok so how can someone force hash collisions?
<Yaniel>
with brute force
<Tjowers>
So they post a lot of data to one of the APIs?
<Yaniel>
try all possible inputs until you find one that has the hash you want
<Yaniel>
usually the hashes are public
<Yaniel>
after all, others have to be able to verify them
<Tjowers>
Ok, so you're saying when database use a hash to shard data among servers, a user can do a DB DoS by flooding a hash bucket
<Yaniel>
no I'm saying you can look up the hash of whatever item is interesting to you (or next to an interesting item)
<Tjowers>
As in by knowing which hash input will go to which server they can flood the server is that how it's applied
<Yaniel>
and start trying to get the same hash (offline) with an input that you control
<Tjowers>
And so after finding the hash -- how does the user attack it -- or do you mean people working internally?
<Yaniel>
and when you have that input you can pretend you came up with the original input
<Yaniel>
and others will believe it, after all you got the hash everyone has seen
<Yaniel>
then you just need to have a majority of the nodes in the network agree on the new truth and everyone else will discard the old info
<Tjowers>
Is this as a data miner or as a regular user?
<Yaniel>
(sibyl attack)
<Yaniel>
a regular user won't have much motivation to do any of this to begin with
<Tjowers>
Maybe I'll go back to the blockchain tangent I was on a couple months ago -- I have networking and crypto concepts understood well enough just need to understand the internals of blockchains -- any good links for the deep stuff?
<Yaniel>
maybe try some infosec stuff instead?
<Yaniel>
so you can practice coming up with ways to defeat your ideas
<Tjowers>
Right, you're saying the underlying principle is information security and to focus on that would be better?
<Yaniel>
probably, depends on what you are trying to achieve
<Tjowers>
Ultimately I think you're hinting at blockchain being more of a challenge of secure consensus on a network
<Yaniel>
correct
<Yaniel>
after all, global distributed consensus is all they are about
<Tjowers>
I like how you think
<joepie91>
obligatory note that a blockchain is almost never the right solution for a given problem, not even if that problem is of the distributed consensus variety
<Yaniel>
^ what I said earlier :P
<joepie91>
(doesn't mean it's not worth learning about how they work, just a thing to keep in mind)
<Yaniel>
learning how they work and how they are defeated is quite insightful yes
<Tjowers>
So blockchain implementations are based on deprecated ideas?
<Tjowers>
Deprecated as in insecure ideas
<Tjowers>
Security moves so quickly
<Yaniel>
well they do have well-known issues that need to be worked around somehow
<Yaniel>
AFAIK consensus is yet far from being a solved problem
<joepie91>
on a related note, worth learning about CT and the more general concept behind it, as implemented in eg. https://github.com/paragonie/chronicle
<joepie91>
where CT = Certificate Transparency
<joepie91>
Tjowers: current-gen blockchain tech does not deliver on its intended security properties (specifically, resistance to manipulation from a powerful actor)
<Tjowers>
Ok, first thing with consensus is establishing chronological order -- current solution is to have a central server decide chronological order -- is there any attempts at decentralized chronological order?
<samueldr>
(is that when leonardo dicaprio tries to sweet talk you?)
<joepie91>
and that is pretty much an unsolved problem currently, yeah
<joepie91>
samueldr: lol
<joepie91>
nice
<Yaniel>
you can't have a central server
<Yaniel>
it won't be distributed then
<Tjowers>
Yeah, the miners are kind of the timekeepers or have they found another solution that's time-independent?
<Tjowers>
That is, events among different machines must be distinguished with respect to time -- unless they each operate functionally
<Yaniel>
no, the miners are more like the crowd that keeps track of "ground truth"
<Tjowers>
Which they could -- if the machines waited for response from another machine before carrying foreword
<Tjowers>
Ground truth being the master copy?
<Yaniel>
that's why things like huge state-owned mining farms are problematic, they can form a large enough group that they can convince others to believe what they say is the truth
<Yaniel>
there is no master copy in a fully distributed system
<Yaniel>
there is only consensus
<Tjowers>
How is size a factor in the algorithm that determines the master copy
<Tjowers>
What couples there farm
<Tjowers>
To the algorithms it should appear as independent servers
<Yaniel>
they do
<Yaniel>
but let's say someone controls 75% of all "independent servers"
<Tjowers>
I suppose the cluster is holds the truth -- and a cluster can be any number of machines which is perhaps where issues arise
<Yaniel>
and they all say that the answer to "tuesday" is "red"
<Tjowers>
I'm listening
<Yaniel>
but the remaining 25% have computed that "tuesday" is "yellow"
<Yaniel>
how do you decide which is true?
<Tjowers>
Ah! because they haven't established chronological order!
<Tjowers>
Because there is no central cluster
<Tjowers>
There's currently no solution to establishing chronological order of events in a distributed system without a centralized server/cluster
<Yaniel>
well the entire idea was to not have any thing "central"
<Yaniel>
because then whoever controls that one central point controls what is considered The Truth
<Tjowers>
And so there are two directions -- solve the "central server tyranny" or the "distributed system oligarchy"
<Yaniel>
and if you decide truthfulness based on chronoloical order alone that doesn't work out either, since you'd just have to be the fastest to come up with $answer
<Tjowers>
decentralized* not distributed
<Yaniel>
which again can be done by throwing enough money at the problem
<Tjowers>
Hm.. so we might need to back up a lot more
<Tjowers>
We ultimately want what?
<Tjowers>
We want secure transfer of information on a network
<gchristensen>
maybe we should have a #nixos-discusses-distributed-computing
<Yaniel>
is that such a common topic xD
<Tjowers>
Essentially -- blockchain aims to create safe exchange of information -- that's literally the internet -- so by building a blockchain we are essentially building a system analogous to DNS
<Yaniel>
secure transfer of information is doable in theory, assuming you can exchange secrets securely (trivial case: face to face)
<Tjowers>
It's essentially an ISP and DNS service
<Yaniel>
actually in practice too, just xor your bytes with a key that is as long as your data
<gchristensen>
Yaniel: mostly that it is a really in-depth conversation which gets really theoretical and isn't exactly -chat :)
<Tjowers>
Blockchain is really a concept that wraps over the concept of the Internet of Things and says "this document/information is the most current, guarantee "
<Yaniel>
hand over the key when you meet the recipient and never use it again, bam, done
<Tjowers>
How is your ISP different from a crypto miner? There are differences, what are the exact differences
<Tjowers>
Actually DNS might be better
<Tjowers>
How is DNS any different from blockchain
<gchristensen>
I can go to court and have DNS things change
<Yaniel>
well yes, DNS answers to the question "what ip has the domain address x"
<Tjowers>
Lmao gchristensen++
<{^_^}>
gchristensen's karma got increased to 198
<Yaniel>
most blockchains answer to the question "what money units does wallet x have"
<Yaniel>
with a traditional bank you can also go to court and change the latter :P
<gchristensen>
thank goodness
<Tjowers>
Brilliant -- best seller material right there man `Blockchains' answer the question "What money units does wallet X have"`
<Tjowers>
That shivered my brain with understanding
<Yaniel>
not intrinsically, but in practice the popular ones answer that question
<Yaniel>
i.e. bitcoin and all its derivatives
<Tjowers>
Yeah it's a distributed hash datastructure
<Yaniel>
namely a structure of... financial transactions
<Tjowers>
It's like if this was a coding interview question "Write a distributed hash tree"
<Tjowers>
Does it hold balances or does it hold the two user's ids along with the net transaction (like +3 and -3 from user 1 and 2 respectively)
<Yaniel>
and as long as everyone agrees which transactions took place (which blockchains try to solve) they can agree on how much currency a given user has
<Yaniel>
idk, haven't checked
<Yaniel>
but it might as well contain both
<Tjowers>
So it's essentially a diff
<Tjowers>
Of the two balances
<Tjowers>
Held in a node
<Tjowers>
Found by the hash of... the hash of what?
<Tjowers>
What is hashed to identify a transaction
<Yaniel>
the hash of the node once it is committed to the chain of other nodes
<joepie91>
Tjowers: it holds neither a balance nor mutations
<joepie91>
at least not in the literal sense
<Tjowers>
What relates a node to another node -- is it that a particular user will have a chain of nodes associated with their transaction history
<Tjowers>
So it holds the current balance
<Tjowers>
Of the user
<joepie91>
actually nevermind
<Tjowers>
And the chain of nodes is the users balance history maybe
<joepie91>
I need to sleep, it's too late for me to start about the nuances of this now :D
<Yaniel>
no, "the blockchain" is a chain of nodes that contain every transaction ever
<Yaniel>
of all users ever
<Tjowers>
I'm sure there are areas where it is implementation specific -- UB
<Tjowers>
So is it one large chain or multiple chains
<Tjowers>
Ah a merkle tree
<Yaniel>
one chain
<Tjowers>
A hash tree
<Yaniel>
no, it's not a tree
<Tjowers>
Ok so the chain never contains two child nodes
<Yaniel>
just a singly linked list
<Tjowers>
No node *
<Tjowers>
Wow, so where does the merkle tree come into play
<Yaniel>
when you have to unify two valid chains I guess
<Tjowers>
Ah for merging
<Tjowers>
Merging truth
<Yaniel>
everyone can't be talking to everyone else all the time after all
<Tjowers>
So it's a way of queuing changes to the truth
<Yaniel>
netsplits can happen, but should not affect the trutfulness of events on either side of the split
<Tjowers>
A network partition?
<Tjowers>
I like netsplits
<Yaniel>
yes, happens all the time on IRC
<Tjowers>
So there are a number of issues to address -- but edge cases after idea
<Yaniel>
IRC solves the situation by... just not handling it
<Tjowers>
Right lmao
<Yaniel>
and actually channels used to be taken over with the help of netsplits
<Tjowers>
So what if there were a ring of "timekeepers" a set of servers that assigned a GUUID to each user upon making a cryptocard and through which each transaction would go through these time keepers to assign the transaction a GUUID based on time -- what are the issues with that?
<Yaniel>
who selects those timekeepers? and what even *is* time?
<Tjowers>
Yeah, so time is a reference used to index a systems change in state
<gchristensen>
and whose gravity?
<Tjowers>
The timekeepers chosen -- that's a thinker
<Tjowers>
You want to avoid central interest -- there might be arguments for a central regulator -- but assuming bias is always not ideal, it would have to be democratically elected
<Yaniel>
and how do you "index change in the state" of something that to the best of your knowledge does not even exist (netsplit)?
<Tjowers>
There is degrees of confidence and levels of uncertainty that could be factored in
<Yaniel>
basically in a distributed system you should never expect to be looking at more than a small fraction of the whole thing
<Tjowers>
The system would be no means go without dispute
<Yaniel>
there will be power outages, connection issues, nodes disappearing, nodes appearing, nodes reappearing after indefinite time, nodes never reappearing after all, ...
<Tjowers>
Existing monetary system -- everyone goes through the private company The "Federal" Reserve -- in the US and in many other countries now
Tjowers has quit [Remote host closed the connection]
<Yaniel>
and nodes that appear "newly" or after disappearing for a while may have experienced things you haven't
Tjowers has joined #nixos-chat
<Tjowers>
Speak of the devil! I had a partition the moment you mentioned it -- mac died
<Tjowers>
Yes, those are solvable things -- it doesn't need to be perfect, it only needs to be correctable
<Tjowers>
Whether programmatically correctable or correctable by a support staff
<Yaniel>
support staff is centralized too
<Tjowers>
Yeah, so the idea I'm conveying is that a central organization is kind of necessary barring future breakthroughs -- On one hand - the government may rule a decentralized currency causes crime to increase and is therefor a threat to national security
<Tjowers>
Like they could easily pull that card
<Yaniel>
and staff would essentially have the power to "correct" their wages with a couple more zeros at the end or something like that
<Tjowers>
Even when it's decentralized it is likely to fall to the influence of large actors
<Yaniel>
yes, some have already ruled such ways IIRC
<Tjowers>
Yes same with the private company The "Federal" Reserve -- they can just change the amount of currency in the United States at any point without permission from our government
<Tjowers>
In fact -- Our country owes the Federal Reserve for the distribution and maintenance of monetary policy
<Tjowers>
They provide a service to the US, and many other countries
<Tjowers>
There is nothing "Federal" about The Federal Reserve -- at least nothing in our Federal government
<Tjowers>
They are a private company -- some of our founding fathers warned us about centralized banks in the beginning -- they got signed in around 1912 or somewhere around there
<Tjowers>
Perhaps centralized banks are kind of useful
<Tjowers>
But we need to get the right people behind monetary policy -- an open community that's transparent, democratic
<Tjowers>
So, how would that work? That's quite a question, but worth asking
<Yaniel>
banks are sort of decentralized though
<Tjowers>
After the Federal Reserve yeah BoA Wells Fargo etc the Visa network
<Yaniel>
they have to agree with each other in order for you to make transactions between accounts in different banks
<Yaniel>
globally
<Tjowers>
Yeah so I sold processor equipment to small businesses I can tell you all about the credit card processing industry
<Tjowers>
To an extent
<Tjowers>
You swipe your card at Walmart -- Walmart pings the bank they designated as the "acquirer"
<Tjowers>
They check to see what network it belongs to
<Tjowers>
Ah, this is a Visa card -- let me look up the issuing bank in their database
<Yaniel>
credit card processing is not really related to banks agreeing on what transactions occured between them though
<Tjowers>
Or they send it to Visa and Visa handles it from there -- I don't think the banks ever communicate directly though it's not clearly stated online
<Tjowers>
The banking network like Visa (I believe) has the master truth
<Tjowers>
It's all batched at the end of the night -- flushed to the proper banks
<Tjowers>
The truth for a given network of banks -- then you have mastercard discover etc
<Tjowers>
They are the miners
<Tjowers>
Yep just googled to confirm -- swipe at walmart, walmart pings their bank, their bank pings visa, visa pings the customers bank make sure they have money, visa lets walmart's bank know the customer has the money and takes note of the transaction, walmart's bank let's walmart know, the little screen says approved.
<Tjowers>
Then you'll have a pending charge until they batch all the updates to balances between banks in the banking network at the end of the night
<Yaniel>
yes but at the end of the day every other bank also has to agree that this transaction took place