/biz/ - Business & Finance

How are you planning to interact with the API onchain without an oracle?

Oh wait you fucking can't, the API to chain Bridge would be the oracle chainlink is just a standardized protocol written very well on how to actually implement this into your contract and make it secure via a large network node if needed.

>>21135653
>Whats the point? Why not just directly access the API?
kek. let me guess, you didnt even know what a smart contract was 2 months ago?

>>21135760
Whats the point? If someone wants to send crypto based on the results of an API, just write a script to run crypto based on the results of an API!

Why do i need to write a (((smart contract))) with some middleman network?

>>21135797
Ive been a professional software engineer for over ten years. I just dont get the point of a smart contract that relies on an external API. Youre just making more steps (and potentially more points of failure) by making a smart contract read the API, when you could just do it yourself.

>>21135869
>tfw you thought you were smart because you’re a “professional software engineer” and now are learning you’re not actually smart at all
It’s honestly a simple concept. Either read more or trust the big brains around you. Or don’t. Nobody here cares about you.

>>21135810
are you still in 2017?

>>21135810
>>21135869
Have you even remotely considered the fact that using an open source smart contract to do all of this completely removes all ability of tampering with the functions of said smart contract?
Why the fuck would i trust some guy to properly run his script and perform the appropriate actions instead of literally just letting computer code do it for me, where it can't lie, it can't cheat, it can't scam.

i guess i shouldnt be shocked that every reply pretends to be from some pseudo-intellectual, but actually explains nothing

>>21135989
Are you stupid? The API can still lie, cheat, and scam. Its all still just code.

>>21135869
>Ive been a professional software engineer for over ten years.
copying html code from the internet doesnt make you a software engineer
> I just dont get the point of a smart contract that relies on an external API.
where does the smart contract get the information it needs from if not from an extrnal api?
> Youre just making more steps (and potentially more points of failure) by making a smart contract read the API, when you could just do it yourself.
who is you and what is he doing? manually type in the result of the api?

>>21136029
A single API can, yes. A decentralized Oracle network that uses an infinite number of APIs reduces the possibility of this to almost 0.

If you rely on 1 API to handle $100M worth of value transfer, it can be compromised. If you rely on 300 APIs, and aggregate the results of those, the chances of it being fed false data reduce dramatically.

>>21135653
3 years

>>21135653
Because at that point you might as well not use crypto as a currency. By trusting a centralized API, you undermine the entire point of using crypto in the first place.

>>21136050
By do it yourself, clearly i mean writing out a script to do it. You know, like how its done now.

>>21136060
If you do it yourself, you can also just query multiple APIs manually. Having some middleman network adds no value in this regard either.

>>21136071
The ((decentralized)) network still needs to rely on potentially (and likely) centralized APIs! The whitepaper for chainlink even specifically gives an example of trusting a flight log API. Why shouldnt i just access that api myself, for free, instantly?

>>21136131
why would i trust you then not to tamper with the data?

>>21136171
Youre already trusting an API to not tamper with the data. A smart contract reads from the same API anyone else can read from.

>>21136131
>If you do it yourself, you can also just query multiple APIs manually. Having some middleman network adds no value in this regard either.

So you are going to set up a script that queries a literal infinite amount of APIs and aggregates those results? And you're going to do this for every single possible piece of external data that you need?
And you are going to force every single person to trust that your source code is bulletproof, that you never modify any of the data?

Do you not see how using a smart contract that can never be manipulated, that can be verified by any third party, is infinitely superior?

Imagine your contract requires data regarding an individual's credit history.
You can either:
1) Write your own custom oracle that aggregates data between multiple credit history providers
2) Pay a reputable external oracle which follows "industry standard" ISO-900X to obtain credit history.

It's clear to me that 2) will win more trust from customers and enterprises.

>>21136203
Let me address this specifically. What you are saying is that with either solution, we have to trust that the API isn't going to feed false data.

This is true.
But with YOUR proposed solution, we have to both trust that:
1. The API isn't going to provide false data
2. YOU aren't going to manipulate true data that the API delivers

There is an entire extra middleman in your process that we have to trust.
Either way, we have to trust that APIs in general are trustworthy.
With YOUR solution we have to additionally trust that you aren't manipulating the true data that the API provides you.

With chainlink, we do not.

>>21136131
This is like saying:
“Crypto relies on centralized miners! Why not just trust one random guy to handle the blockchain?“

>>21136211
Do you understand how a smart contract is made? Its still just code that also needs to define the APIs it has to access.

The only difference is the language and which PC is running the API queries. Do you actually look over the code of every smart contract? Do users? No, of course not. Just like how you and nobody else has actually read through every line of the OS you use. At a certain point, you have to trust somebody, just like both a smart contract and not both have to trust the API they want to query.

>>21136226
Both are querying the same API. You could standardize an output based on the aggregate without a smart contract network being a middleman. In fact, it already works like this! How do you think Discover, Visa, or your landlord know your credit score?!

>>21136308
Not true. Both parties can run the same API query to automatically send a payment. Then you dont need to trust anyone else. Just yourself and the API.

>>21136311
False equivalence. Mining serves a clear purpose, which is mathematically ensuring the validity of transactions.

Oracles won't be needed in the future, maybe link can reinvent itself

>>21136360
>Not true. Both parties can run the same API query to automatically send a payment. Then you dont need to trust anyone else. Just yourself and the API.
What do you even mean by this?

Who gets to determine what/where the payment is sent? Who?

If I have a contract that says "If the S&P 500 is $4000 or higher on December 31st 2020, pay Jake $1,000,000"

Who exactly gets to be the ultimate deciding factor on whether or not Jake gets his $1M?
Jake runs an API query, and YOU run an API query?

And what if you both have different results? Who gets to determine the ultimate outcome of this situation.

>>21135653
>>21135810
>>21135869
>Why not just use HTTP instead of HTTPS?

>>21136414
Right now? A stock broker. Any smuck could open up Robinbood and do this exact same thing without a smart contract. The only difference is the broker is now the one doing the API queries.

>>21136521
A centralized entity. Nice.
And if I want to make the same bet but not give a fee cut to a centralized entity? Oops guess its not possible

>>21136510
What a stupid post. False equivalence with zero self awareness. SSL serves a purpose, which is that it encrypts data transfer between you and a server, so a middleman cannot listen in or tamper with the data.

>>21136360
But, it’s worthless without a decentralized network of miners which can produce a deterministic result.

Since my shitposting clearly isn’t cutting it, if you want to research this more for yourself, google “the oracle problem”. It will help you understand why a crypto-secured decentralized network is undermined by non-crypto-secured centralized point.

I mean, ffs, go write a smart contract in solidity and try to query an API. You can’t. It would destroy the blockchain entirely.

>>21136549
Chainlink isnt free, you are still paying a fee to somebody for this api query.

You arent escaping centralization with smart contracts, because both a broker and a smart contract network are asking the same API(s) for the price of that stock.

>>21136558
>so a middleman cannot listen in or tamper with the data.
So close, yet so far

No they are not querying the same API. I gave a HYPOTHETICAL example. You can either build your own oracle to do the calculation, or pay an external oracle which follows the standards of an industry agreed ISO-XXXX standard to aggregate the data.

>>21136521
So a middleman? The thing you were complaing about?

>>21136601
>You arent escaping centralization with smart contracts, because both a broker and a smart contract network are asking the same API(s) for the price of that stock.
You are absolutely escaping centralization.
Literally in the previous post I explained that there must be a party in your scenario who gets the ultimate control of how to send that $1,000,000.

With smart contracts, with blockchain, with chainlink, there is no single party who gets to press the big red button "Send Jake his $1M"

>>21136614
>>21136622
My point is a smart contract network is just replacing one middleman with a much bigger and more complicated middleman, and for what gain? Everyone accessing the same API, so there isnt really any added security, just the illusion of it

>>21136558
>between you and a server, so a middleman cannot listen in or tamper with the data.
It's exactly this but for smart contracts and APIs, also look up Mixicles.

>>21136638
But there is! The one who hosts the API! Sure there may be multiple to query, but do you really believe they are all owned and controlled by different people??

>>21136004
I mean if you cannot figure out why a centralized Source of data is bad for a smart contract, why are you even here?

>>21136060
Not just aggregate the data, but penalize them for providing false data and incentivize them for providing good data!!!

>>21136720
Smart contracts are still accessing a centralized source of data! The API they eventually use to determine the result of the contract

>>21136711
>>21136759
Again. With your solution, we have to not only rely on the API to feed true data.
BUT we also have to rely on the centralized entity who is reading that API to then send the money in an appropriate manner with regards to the data that is pulled from that API.

With chainlink, the API(s) are the single source of all power.

>>21136808
Whats the point if you still need to trust the API?

>>21136825
................................................ to not have to trust the centralized entity to act accordingly to the data from the API
it completely removes an entire layer of trust

>>21136849
I get what youre saying, anon. I really do. In the cases of both parties entering a contract and relying on a centralized external api neither of them control, it seems mildly useful because you can remove a single layer of trust (the other party in the contract).

The thing that gets me the most though, is the example in the whitepaper. Their example is, enter into a contract with an airline. If their api says your flight made it, the contract will send your money to them. I think this is a more real world example of what people would blindly use smart contracts for. Just blindly trusting the api that one party entirely controls. In these cases, chainlink adds zero benefit.

I bring this up because its in the whitepaper, which leads me to believe that even the devs know that this is what it'll really be used for.

>>21136992
One of the largest advantages to using blockchain or smart contracts in general is settlement time.
The entire financial industry as we know it today is slow as fuck. If you use smart contracts along with blockchain, you can reduce the time to payment for certain financial instruments (derivatives, etc) to mere minutes, instead of days.

This is the power of using blockchain. Money is transferred in real time with no central authority having the keys to the castle.
In order to use blockchain on a scale like this, you need tamper-proof contracts, aka smart contracts.
In order to use smart contracts in this manner, you require a way to feed data into those contracts. This is where Oracles come in.

>>21137088
Im already sold on crypto in general. I agree, yeah, the financial system as a whole sucks. Its a few big players who hold the keys to everything and transaction speeds are slow.

I dont see this related to smart contracts though. In terms of "if API replies with X, send Y to Z" can also be done with crypto both with and without smart contracts.

Im really trying to see the value of smart contracts here, or else i wouldn't have made this thread.

So far ive been convinced that gambling is a pretty good use case for smart contracts, but nothing else yet.

>>21137214
If you don't use smart contracts, you do not take the keys from the people who control everything.
They still hold absolute and complete control over when and where the transaction is sent.

>>21135810
>If someone wants to send crypto based on the results of an API, just write a script to run crypto based on the results of an API!
Yep, creating an oracle is as simple as wiring a python script that queries the API, then sends a transaction with the data using a library like web3.py, then running the entire thing on a cheap VPS or Amazon cloud instance.

You pay nothing but gas, no extra token required.
This is how to do a centralized oracle.

The only "advantage" of Chainlink is it is supposedly "decentralized".
Nevermind that
- the source of the data itself is centralized
- and that their "solution" to decentralization is KYCing the nodes.

Decentralization of oracles (which is an unsolved computer science problem) was not solved in the whitepaper and is merely an afterthought, despite being the main selling point.

Imagine if bitcoin was released by Satoshi, without having solved the Byzantine generals problem. And his solution to the double spending problem is KYC nodes. No PoW. Just KYC. Slapped together with some sort of "penalty system".

This is what Chainlink is doing.

But why is it $11 and rising still?
Because memes anon. Memes got the orange man the presidency. It's just that powerful.
That's really all that matters.

>>21137472
But sergey is Satoshi.

>>21137472
You do a fantastic job at undermining the absolute extreme importance of the penalty system.
It makes the KYC nodes issue evaporate, and ensures that all nodes have zero financial incentive to act dishonestly.

>>21137214
props to this guy he tried really hard to depate the autists who call themselfs the link marines.

FUCK FINALLY A REAL THREAD ABOUT LINK

I mean OP is obviously too fucking stupid to just read the whitepaper and understand why direct API access is profoundly stupid

But still BLESS THE ANONS IN THIS THREAD POSTING REAL SHIT

>>21135653
You had 3 years my man, it's a little bit late to be playing this game now.

>>21137472
It does feel like they've really tricked people into thinking they're intelligent by using big boi words like "smart contract." Its like they forgot APIs the contracts depend on are centralized and think they're geniuses for ignoring that fact or pretending its irrelevant. Just look at this thread.

>>21137726
There's nothing wrong with direct API access. What a stupid assertion.

>>21137757
>Smart contract is a big boi word

well im out, OP is clearly just trolling. he is back to saying things that he has already conceited to be incorrect.
oh well, back to staring at my $150,000 grow hilariously fast

>>21137802
what if u need to access more that one api? for ex. the flight data question could be solved with lets say 5 different flight apis connected to smart contract etc.

>>21137855
conceded* kek, im a brainlet. anyways have fun with your thread boss man.

The best solution: The API provider writes their data to a secret contract, then other people consume it and pays the API provider, if the API provider fuck something up then hey lose their SCRT stake. Then you can have other secret contracts that examine the data without revealing the contents.

Chainlink is just a bandaid solution after all

https://blog.scrt.network/

>>21137802
you’re not as smart as you think

>>21137879
Then make your shitty python or c# script access all of them? Its not hard

>>21137802
ARE YOU LITERALLY HELMET WEARING RETARDED
DO YOU KNOW ANYTHING ABOUT THE HISTORY OF DEFI
DO YOU THINK YOU KNOW MORE ABOUT THIS THAN ALL OF THE LEADING DEFI TEAMS THAT USE LINK?

Your arrogance is why people like me treat faggot software engineers like you like the cuck slaves you are

Try actually thinking with your brain for once rather than your faggot cuck feelings

>>21137908
You're one to talk. You made an idiotic statement and the continue to bark like a dog, and yet still haven't provided a single fact which would back up your initial statement.

>>21137928
why dont u do this then? prove everyone that u are right and link is uselss. u might be able to short link meanwhile.

>>21135869
It would just automate it technically dont need link to do it could just make a standard code that everyone uses and trusts.

>>21137989
The whole world is already doing this, anon. You're doing it right now by posting on 4chan by your browser accessing their simple API. You're even accessing googles API for every captia you fill out.

When you have a ticker on your phone as a widget? Usually they access multiple APIs. This is how it works everywhere already

>>21135810
This is the equivalent of saying why should every company use the phone network when they can build their own telecommunication network. Why should every company use windows when they can build their own OS. Why should should every company use volvo trucks when they can build their own transportation?

>>21138161
False equivalence. The level of effort of writing an api query is so low. A literal child could figure it out.

>>21138219
>$11.46

>>21138492
>11.75

>>21138219
Anon, most people don't even know what an API is. I work for an e-commerce platform and it's an absolute pain in the ass to get customers (that are in tech too) to understand how an API works.

>YOUR PLATFORM IS BROKEN. I CANT SEE THE ORDERS IN MY SYSTEM
>Hi, you're currently fetching orders through our API. If you have trouble fetching orders, please check your call. If you are unable to figure it out, please send us the call so we can tell you what is wrong with it.
>"N-NO! I NEED MY ORDERS NOW SO I CAN SHIP. KINDLY EXPORT THE ORDERS TO US RIGHT NOW. WE ARE LOSING MONEY!
>Sir... We can't. You have to use a GET request to our /orders endpoint so you can import the unprocessed orders yourself. We cannot export them to your system.
>PLEASE EXPORT THE ORDERS OR I WILL ESCALATE. THANKS IN ADVANCE.
You would be amazed by how many big players in e-commerce have senior level staff that do this.

>>21138063
i’m only spoonfeeding because i see a lot of other well meaning anons who don’t quite get it either.
do you know how a smart contract works? this isn’t a stupid perl script; whatever code you push to ethereum has to be run by every node and has to return the SAME RESULT. every node has to call the API but they aren’t all going to do it at the same time which means if the api results change in that time they won’t all have the SAME RESULT.

an oracle sits offchain and makes the api call and locks in the results in a transaction onchain so that when your smart contract makes a call it’s to the onchain transaction and not the API directly. this ensure that each node has the same result.

and yes, of course you can make your own oracle, but obviously it becomes a big vulnerability, data integrity, availability, etc. are all issues. this makes chainlink is the centralized point of failure for now but they SPECIALIZE in this which is why people use it, because they will do it BETTER than you would. eventually when staking is live and the full network is built out, it won’t be a single point of failure anymore.

and yes, if the api is corrupted or unavailable then there’s nothing that can be done for it. chainlink can’t make any guarantees for the api itself (no one can) all it guarantees is that whatever data the api provides is what appears onchain right or wrong.

>>21137564
How do you implement a penalty system? You do realize this is a very difficult statistical problem?

>>21137726
You sound very low IQ

>>21139737
>>21139919
You sound like a typical midwit.
>Muh difficult statistical problem!
No shit, why do you think the space is going crazy about LINK and their proposed solution?

>>21139170
Do you think this thread is just an example of highly skeptical people not having a complete grasp of the problem and tech and therefore never being able to be satisfied enough with any explanation?

>>21140173
Definitely. It's not really a bad thing to be skeptical, threads like this used to be very common in 2018. Eventually anons saw the light and grasped the concept. Some take longer than others. However, anons like this >>21139737 that just spout dumb rhetorical questions to signal how intelligent they themselves are, are never going to make it. Imagine not even glossing over the whitepaper before spewing nonsense like this.

>>21135653
i agree with you, companies that really want this could do this on their own without much effort and it'd cost them a lot less than using link or band or whatever
i still hold link though because companies want to see where this tech goes and if it's any use at all for reselling, like oracle is doing
i'm in it for the money strictly, i understand that the tech isn't needed
it is however good for companies that don't have the means to develop something specifically suited for their needs and would rather pay a premium for an oracle service like link, band, etc.
there are plus sides to using decentralized oracles that come along with bringing off-chain data on-chain
so as for your questions
>Whats the point?
for companies to try it and see if it's worthwhile, time will tell if there is an actual demand for this long-term
>Why not just directly access the API?
you could, but you'd have to basically build this, which is easy for most companies that even want something like this, but for now one exists and can be tested to see if it's even worthwhile. if it is, they'll go inhouse with it
>Seems like a useless middleman.
i wouldn't say useless, unless we're talking about how much it'd cost for this service, then it does seem useless
>A lot of abstraction, for what gain?
link specifically went the jack of all trades route with their service, so that's where all the abstraction comes from. as for companies integrating with chainlink, the gain right now is to gauge the usefulness of a decentralized oracle service before they spend money building something for themselves that in the long run would be much cheaper

in the end, it's all hype brother, if u got in at a cheap price, u would have good unrealized gains and depending on how much further u think the hype would take it, u'd hold or take profit
i'm currently still holding, will probably take profit around $12.50-13, but that all depends, i mean if it blows through $12 like it did $11 today, i'll hold out

>>21140520
you're retarded too

>>21140500
One of the advantages both retarded and high IQ people understand is that their intelligence has limitations. Midwits typically suffer from the dunning-kruger effect and refuse to let go.

>>21136664
The "new middleman" is automated, instant and tamper-proof.
If you can't see the value of this, you might be in the wrong business .

>>21140126
>>21140500
How is what I’m asking a dumb question? If the output of the aggregator is a median (for example) which responses are penalized? Responses k * std from the median?

>>21140939
It's not a dumb question. You're just dumb for not figuring out where to find the answer.

>>21140993
Where is the answer? I’ve read the white paper unironically and have not seen anything other than vague mention of penalization and outlier removal.

Is the answer “it’s configurable”? The standard cop-out answer CL always gives?

>>21139623
But APIs already have timestamps in their output data that often go into the millisecond that makes validating many API calls easy. I can see the usecase where perhaps an API doesn't have the functionality to output historical data past a certain point in time where verified data stored on chain that was pulled in the past would be useful, but real time API verification doesn't seem as revolutionary as chainlink holders make it seem. Even if you were trying to verify multiple different API calls for the same data, like say for example stock data - every broker/API source has slightly different data for the same timestamp. You can see this directly by observing price action on multiple broker platforms and even services like Yahoo finance. This obviously makes verifying data integrity a nightmare.

>>21140520
This seems like a more reasonable take on chainlink, I can perhaps see smaller companies utilizing some functionality of chainlink but I think a lot of bigger firms prefer to be in absolute control of their data and have the resources to make that to happen.

>>21141239
>Is the answer “it’s configurable”? The standard cop-out answer CL always gives?

tfw "chainlink is extremely flexible and can be used for an infinite number of situations in an infinite number of ways to meet unique demands of anyone wanting to use the network" is FUD

>>21136992
Think about the big short. Remember when their bet on the housing market was becoming profitable, but the banks (brokers) in your examples wouldn’t pay out? They kept saying their metrics didn’t support paying out when clearly they were manipulating it and waiting to unload bags themselves first

>>21141239
>can configure the parameters of risk to suit your needs
>th-that's just a cop-out!!
thank you for the entertainment
HOLY FUCKING KEK

>>21141239
>>/biz/

>>21141303
You’re implicitly assuming the specific configuration is easy to find and even exists. Automated outlier removal is a very difficult problem (and I think there is a lack of statisticians in the crypto space, so a lot of people don’t know this). Making your clients figure this out for themselves doesn’t sound like a good approach.

>>21141415
Further, how do you know an outlier is not a response that is outperforming all the other responses rather that underperforming?

>>21141415
>>21141490
A man who literally invented proof of work (Ari Juels) is behind chainlink.
People more intelligent than you can possibly imagine are tackling this. Give it up, anon.

>>21141521
A good computer scientist does not imply a good statistician. This is actually quite rarely the case since computer scientists exclusively think about determinism and statistics is about non determinism. The exception being machine learning researchers.

>>21135810
>>21135989
You guys are slowly understanding the limited usability of smart contracts. Clearing will not be simply replaced by some self executing lines of code. It would take negotiations and lawyers to create this between companies. Smart contracts are good for non-complex issues like tracking shipments, however here i don't really need a smart contract. Simple company interfaces or programs can perform that without paying fees to node operators for solving cryptographic tasks. People in here fail to assess sc usability due to a complete lack of business experience.

>>21141601
I'm done replying to you, if you truly think that you have discovered a fundamental issue with chainlink that they haven't already deeply considered and made plans to resolve

Then you are the most egotistical narcissistic person to post ITT.
You are not more intelligent than Ari Juels or Steve Ellis. They have done things in years that you couldn't accomplish in centuries.

>>21141736
every week we get these d.krug chuds coming in thinking they’re asking things no one else has before. Like I get not understanding the problem or its solution and asking questions to figure it out, but invariably these chodes always get pissy and hyperaggressive in less than ten minutes of discussion. It’s so tedious

guys, OP is correct and the entire defi space using oracles are wrong. someone call them and let them know they can just query 3-4 different APIs themselves and problem solved

>>21141736
>appeal to authority argument
Truly a cult

For a software engineer, you sure work really stupid.

>Be software engineer
>Have choice:
>Create 12 adapters for 12 different APIs
>Create generic adapter that I'd have to flesh out extra connections to & be responsible for
>Create 1 Chainlink adapter, they continually add new connections and are responsible for it

>>21141981
The current best solution does not imply is doesn’t have flaws nor that a better solution will be created.

>>21141736
>You are not more intelligent than Ari Juels or Steve Ellis.
Intelligence isn’t a one-dimensional metric. One can be very smart about a particular domain but not another. E.g. Sergey is clearly very intelligent but I doubt he could solve a Leetcode medium.

Tell me, if chainlink has thought deeply about the statistical issues of their system, why haven’t they hired any statisticians?