Steem pyramid scheme revealed

[iamnotback]

Censorship still requires a nuclear option, no matter who pushes the button.

So could Steemit be destroyed if someone posted a bunch of "illegal content" on there? If so, that's a HUGE flaw in a business plan considering the current state of The State.

How does Steemit deal with the Digital Millennium Copyright Act, or, creepy kiddie touchers posting links and pix. I know pix referenced on Steemit are hosted elsewhere, but if references to them are hardwired into the blockchain, and it would take a hard fork to remove them.....

Seems like those two things right there doom this to fail.

We mentioned this already upthread (in one of the Steem threads).

I think the solution is do not host the data on the blockchain. Host only the hash of the data.

The data should be served by other nodes. These nodes will decide which content they feel legally safe to serve.

Other technologies such as Bittorrent, IPFS, etc can attempt to serve this content as well.

The data will never be entirely deleted, it will just be an issue of whether it is easy to locate.

Users could digitally sign certifications of ownership or attribution of the source of the content, which would give nodes legal standing to host the content. Users then fight with copyright infringement claims. Perhaps registries will emerge.

[iamnotback]

The data I need which I think will show if Steem is growing virally or not:

But you have not answered the most important part:

Lets assume for a second you are right, how long to you expect it will take until we will see Steemit breaking down? Would you be open to changing your mind if it doesn't? Do you see this as inevitable, or do you think there is any important step which if taken could prevent the "implosion"?

What are your thoughts on that?

I am watching for either daily or monthly uniques not growing according to the logistic function adoption of new technologies. The monthly uniques are not reported and I don't know if anyone is recording this data periodically so they can chart it. Or otherwise building blockchain analysis tool to extract the data.

Also some way to differentiate between bot (including duplicate Sybil) users and real human users w.r.t. to that data.

Also I'd like some stats on demographics, especially on females and those coming who had no interest in crypto before they arrived.

[smooth]

Censorship still requires a nuclear option, no matter who pushes the button.

So could Steemit be destroyed if someone posted a bunch of "illegal content" on there? If so, that's a HUGE flaw in a business plan considering the current state of The State.

This is little different from any other blockchain. There are links to CP and probably some other bad stuff mined into Bitcoin's blockchain.

Links are perhaps the least problematic since there is always someone else you can go to to try to get the data taken down.

It is very hard to prevent people from storing data on a blockchain. Even if you just store hashes, transactions, signatures, etc. all can be used/abused/misused (take your pick) to store data.

[generalizethis]

...but, if I can do it, I'll do it within the confines of 14 lines, a nuanced rhythm, and with a few pics--no promises on how it turns out--gibberish or nein (but I promise ya neither theo now.

I'll just admit I am a literary dunce and can't decipher your poetry.

I think you'll like this (binary turns are fun--and it's topical): https://steemit.com/olympics/@generalizethis/a-competitor-s-plea-to-phelps

[MWD64]

Censorship still requires a nuclear option, no matter who pushes the button.

So could Steemit be destroyed if someone posted a bunch of "illegal content" on there? If so, that's a HUGE flaw in a business plan considering the current state of The State.

This is little different from any other blockchain. There are links to CP and probably some other bad stuff mined into Bitcoin's blockchain.

Links are perhaps the least problematic since there is always someone else you can go to to try to get the data taken down.

It is very hard to prevent people from storing data on a blockchain. Even if you just store hashes, transactions, signatures, etc. all can be used/abused/misused (take your pick) to store data.

What about copyrighted material (the actual text, not links) being posted on Steem? What if someone posted chapters of a Harry Potter book, or something else from a corporation that is heavily litigious?

How would Steem deal with a DMCA demand?

What about if someone doxxed government officials on Steem?

What if someone posted Daniel Larimer's home address?

What if someone posted some company's "trade secrets?"

Or started using it as the new wikileaks to post "stolen" "government secrets"?

I'm just curious about how "a site that can't be censored" (without a hard fork) deals with the eventuality of when government guns come HARD and say "you MUST delete the following things....."?

I hate intellectual property laws but this is stuff that a company has to deal with if they have a central location for offices and servers, and have real people's names on corporate papers and can be shut down.

If Steem ever got 1/1000th as many users as Facebook, there would have to be an hourly hard fork. Think of all the stuff that Facebook or  Wordpress.com (the blog hosting part) HAVE to delete daily to stay in business. How will Steem deal with that?

Business plan: someone should make a blogging version of Open Bazaar. Would be WAY more useful than Steem from the censorship end.

[iamnotback]

Censorship still requires a nuclear option, no matter who pushes the button.

So could Steemit be destroyed if someone posted a bunch of "illegal content" on there? If so, that's a HUGE flaw in a business plan considering the current state of The State.

This is little different from any other blockchain. There are links to CP and probably some other bad stuff mined into Bitcoin's blockchain.

Links are perhaps the least problematic since there is always someone else you can go to to try to get the data taken down.

It is very hard to prevent people from storing data on a blockchain. Even if you just store hashes, transactions, signatures, etc. all can be used/abused/misused (take your pick) to store data.

Technically incorrect in the case of asymmetric public-key cryptography signatures.

Afaik it is mathematically intractable to find a public key that enables you to produce a signature that maps to some chosen data.

So then we don't store the hash of what was signed. The data provider provides the match. So the blockchain doesn't store it.

Also there may be a heuristic to detect if a hash is not randomized data.

[smooth]

Censorship still requires a nuclear option, no matter who pushes the button.

So could Steemit be destroyed if someone posted a bunch of "illegal content" on there? If so, that's a HUGE flaw in a business plan considering the current state of The State.

This is little different from any other blockchain. There are links to CP and probably some other bad stuff mined into Bitcoin's blockchain.

Links are perhaps the least problematic since there is always someone else you can go to to try to get the data taken down.

It is very hard to prevent people from storing data on a blockchain. Even if you just store hashes, transactions, signatures, etc. all can be used/abused/misused (take your pick) to store data.

Technically incorrect in the case of asymmetric public-key cryptography signatures.

Afaik it is mathematically intractable to find a public key that enables you to produce a signature that maps to some chosen data.

So then we don't store the hash of what was signed. The data provider provides the match. So the blockchain doesn't store it.

Also there may be some way to detect if a hash is not randomized data.

I didn't say the signature has to match, but it can still encode data.

You can employ countermeasures, They may increase the cost of encoding the data, but for things like links, the amount of data that needs to be shared is pretty low, and the value of sharing them may be very high. This becomes a likely-unwinnable arms race.

What about copyrighted material (the actual text, not links) being posted on Steem? What if someone posted chapters of a Harry Potter book, or something else from a corporation that is heavily litigious?

How would Steem deal with a DMCA demand?

What about if someone doxxed government officials on Steem?

What if someone posted Daniel Larimer's home address?

What if someone posted some company's "trade secrets?"

Or started using it as the new wikileaks to post "stolen" "government secrets"?

I'm just curious about how "a site that can't be censored" (without a hard fork) deals with the eventuality of when government guns come HARD and say "you MUST delete the following things....."?

Good questions. They'll certainly all be removed from the web site. Beyond that we are in uncharted territory for the most part. Bitcoin had CP links mined into it, and probably other illegal numbers. So far that has not become a major problem. Steem might force the issue.

[iamnotback]

Censorship still requires a nuclear option, no matter who pushes the button.

So could Steemit be destroyed if someone posted a bunch of "illegal content" on there? If so, that's a HUGE flaw in a business plan considering the current state of The State.

This is little different from any other blockchain. There are links to CP and probably some other bad stuff mined into Bitcoin's blockchain.

Links are perhaps the least problematic since there is always someone else you can go to to try to get the data taken down.

It is very hard to prevent people from storing data on a blockchain. Even if you just store hashes, transactions, signatures, etc. all can be used/abused/misused (take your pick) to store data.

Technically incorrect in the case of asymmetric public-key cryptography signatures.

Afaik it is mathematically intractable to find a public key that enables you to produce a signature that maps to some chosen data.

So then we don't store the hash of what was signed. The data provider provides the match. So the blockchain doesn't store it.

Also there may be some way to detect if a hash is not randomized data.

I didn't say the signature has to match, but it can still encode data.

The validators will not accept invalid signatures. Thus no way you can store any data you want on the blockchain. What is being signed wouldn't be stored on the blockchain. Please don't pretend you don't understand.

[smooth]

Censorship still requires a nuclear option, no matter who pushes the button.

So could Steemit be destroyed if someone posted a bunch of "illegal content" on there? If so, that's a HUGE flaw in a business plan considering the current state of The State.

This is little different from any other blockchain. There are links to CP and probably some other bad stuff mined into Bitcoin's blockchain.

Links are perhaps the least problematic since there is always someone else you can go to to try to get the data taken down.

It is very hard to prevent people from storing data on a blockchain. Even if you just store hashes, transactions, signatures, etc. all can be used/abused/misused (take your pick) to store data.

Technically incorrect in the case of asymmetric public-key cryptography signatures.

Afaik it is mathematically intractable to find a public key that enables you to produce a signature that maps to some chosen data.

So then we don't store the hash of what was signed. The data provider provides the match. So the blockchain doesn't store it.

Also there may be some way to detect if a hash is not randomized data.

I didn't say the signature has to match, but it can still encode data.

The validators will not accept invalid signatures. Thus no way you can store any data you want on the blockchain. What is being signed wouldn't be stored on the blockchain. Please don't pretend you don't understand.

The sender can change the data slightly which produces a different signature. This can be used to encode whatever data you want. If the data is encrypted or otherwise obfuscated then it is between difficult and mathematically impossible to distinguish it from random.

[iamnotback]

Censorship still requires a nuclear option, no matter who pushes the button.

So could Steemit be destroyed if someone posted a bunch of "illegal content" on there? If so, that's a HUGE flaw in a business plan considering the current state of The State.

This is little different from any other blockchain. There are links to CP and probably some other bad stuff mined into Bitcoin's blockchain.

Links are perhaps the least problematic since there is always someone else you can go to to try to get the data taken down.

It is very hard to prevent people from storing data on a blockchain. Even if you just store hashes, transactions, signatures, etc. all can be used/abused/misused (take your pick) to store data.

Technically incorrect in the case of asymmetric public-key cryptography signatures.

Afaik it is mathematically intractable to find a public key that enables you to produce a signature that maps to some chosen data.

So then we don't store the hash of what was signed. The data provider provides the match. So the blockchain doesn't store it.

Also there may be some way to detect if a hash is not randomized data.

I didn't say the signature has to match, but it can still encode data.

The validators will not accept invalid signatures. Thus no way you can store any data you want on the blockchain. What is being signed wouldn't be stored on the blockchain. Please don't pretend you don't understand.

The sender can change the data slightly which produces a different signature. This can be used to encode whatever data you want. If the data is encrypted or otherwise obfuscated then it is between difficult and mathematically impossible to distinguish it from random.

I already wrote in 3 posts, that the data being signed wouldn't be stored on the blockchain. So only the signature would be stored. Afaik, it is mathematically intractable to find some set of (public key, signed data) that will produce a desired signature.

[smooth]

Censorship still requires a nuclear option, no matter who pushes the button.

So could Steemit be destroyed if someone posted a bunch of "illegal content" on there? If so, that's a HUGE flaw in a business plan considering the current state of The State.

This is little different from any other blockchain. There are links to CP and probably some other bad stuff mined into Bitcoin's blockchain.

Links are perhaps the least problematic since there is always someone else you can go to to try to get the data taken down.

It is very hard to prevent people from storing data on a blockchain. Even if you just store hashes, transactions, signatures, etc. all can be used/abused/misused (take your pick) to store data.

Technically incorrect in the case of asymmetric public-key cryptography signatures.

Afaik it is mathematically intractable to find a public key that enables you to produce a signature that maps to some chosen data.

So then we don't store the hash of what was signed. The data provider provides the match. So the blockchain doesn't store it.

Also there may be some way to detect if a hash is not randomized data.

I didn't say the signature has to match, but it can still encode data.

The validators will not accept invalid signatures. Thus no way you can store any data you want on the blockchain. What is being signed wouldn't be stored on the blockchain. Please don't pretend you don't understand.

The sender can change the data slightly which produces a different signature. This can be used to encode whatever data you want. If the data is encrypted or otherwise obfuscated then it is between difficult and mathematically impossible to distinguish it from random.

I already wrote in 3 posts, that the data being signed wouldn't be stored on the blockchain. So only the signature would be stored. It is mathematically intractable to find some set of (public key, data) that will produce a desired signature.

The signature is the data. Imagine one bit per signature. That's enough to transfer short links without too many signatures (but it isn't hard to do better).

You can easily encode data in the amounts of payments. If amounts are hidden you can encode data in the number of outputs, or if number of outputs is fixed, you can encode data in the timing of payments.

There are a million ways. How contrived these measure get will be a function of how much effort is put into suppressing it.

[iamnotback]

Censorship still requires a nuclear option, no matter who pushes the button.

So could Steemit be destroyed if someone posted a bunch of "illegal content" on there? If so, that's a HUGE flaw in a business plan considering the current state of The State.

This is little different from any other blockchain. There are links to CP and probably some other bad stuff mined into Bitcoin's blockchain.

Links are perhaps the least problematic since there is always someone else you can go to to try to get the data taken down.

It is very hard to prevent people from storing data on a blockchain. Even if you just store hashes, transactions, signatures, etc. all can be used/abused/misused (take your pick) to store data.

Technically incorrect in the case of asymmetric public-key cryptography signatures.

Afaik it is mathematically intractable to find a public key that enables you to produce a signature that maps to some chosen data.

So then we don't store the hash of what was signed. The data provider provides the match. So the blockchain doesn't store it.

Also there may be some way to detect if a hash is not randomized data.

I didn't say the signature has to match, but it can still encode data.

The validators will not accept invalid signatures. Thus no way you can store any data you want on the blockchain. What is being signed wouldn't be stored on the blockchain. Please don't pretend you don't understand.

The sender can change the data slightly which produces a different signature. This can be used to encode whatever data you want. If the data is encrypted or otherwise obfuscated then it is between difficult and mathematically impossible to distinguish it from random.

I already wrote in 3 posts, that the data being signed wouldn't be stored on the blockchain. So only the signature would be stored. It is mathematically intractable to find some set of (public key, data) that will produce a desired signature.

The signature is the data. Imagine one bit per signature. That's enough to transfer short links without too many signatures (but it isn't hard to do better).

I am referring to an ECC signature. You are apparently thinking of transaction data? I don't know why it is so difficult for you to understand that afaik it is mathematically intractable to control the bits of an ECC signature.

You can easily encode data in the amounts of payments. If amounts are hidden you can encode data in the number of outputs, or if number of outputs is fixed, you can encode data in the timing of payments.

There are a million ways. How contrived these measure get will be a function of how much effort is put into suppressing it.

Payments are numbers. Anyone can write any filter they want to reinterpret any data any where on the Internet as anything. Everything would be illegal. So obviously this will never stand up in any sane court-of-law.

You are being absurd.

Are you interested in improving things or just nonsense.

[generalizethis]

You can easily encode data in the amounts of payments. If amounts are hidden you can encode data in the number of outputs, or if number of outputs is fixed, you can encode data in the timing of payments.

There are a million ways. How contrived these measure get will be a function of how much effort is put into suppressing it.

Payments are numbers. Anyone can write any filter they want to reinterpret any data any where on the Internet as anything. Everything would be illegal. So obviously this will never stand up in any sane court-of-law.

You are being absurd.

Are you interested in improving things or just nonsense.

So if I'm reading smooth's comment correctly, couldn't multiple ui's interpret the information differently, allowing for decentralized use? I mean you could go after steem, but that (eventually) turns into a game of whack-o-mole.

(I understand things through words, so forgive me if I'm missing mathematical parts.)

[iamnotback]

I think VESTS are only created or destroyed by power up, power down, or rewards (that pay in SP/VESTS), unless I'm forgetting something.

If you sell SD on an exchange, literally nothing changes in terms of supply, just the owner of the existing token. That seems kind of obvious, no?

Yes changes in the price of STEEM (as reported by the oracles i.e. witnesses) between the time SD is created and when it is destroyed will influence the money supply (virtual supply; converted to real supply if and when the SD->STEEM conversion takes place). If I'm not mistaken that is covered in the white paper.

There is no special tagging of VESTS nor the STEEM/SP in the vesting fund.

I think the confusion was I thought commentators were stating that no liquid STEEM were in the vesting fund and I thought only VESTS would go into a "vesting fund". So I was confused as to where the liquid STEEM were accounted.

Now I think I've been told the VESTS are not in the vesting fund, so therefor liquid STEEM are accounted for in it, and the SP are accounted for in the VESTS. Is this correct? If not please provide some complete definitions. This is  getting ridiculous. How can anyone know WTF is going on if the damn terms and design isn't documented. I try to guess based on different statements which makes it look as though it is my error, but it is a failure of clear communication.

Can you tell me where in the white paper? I don't remember reading about the ability to retire/destroy SBD.

I don't see a particular sentence but the concept is woven throughout the SBD section, for example in describing SBD as analogous to convertible notes, mentioning that conversion can only be done in one direction, etc.

What I understand now is that when we "Convert" our SBD, it retires them and converts to liquid STEEM. Whereas, when we choose "Buy or Sell", then we exchange the SBD for liquid STEEM and the SBD is not destroyed.

Then how do we know which VESTS to transfer to the party who owns the SBD when a SBD is retired? Or what does it mean to destroy a SBD?

First of all converting SBD does not deliver VESTS, it delivers liquid STEEM (which can then be converted to VESTS via a power up). Second of all, VESTS are fungible so there is no such thing as "which VESTS".

Again it was the commentators in my blog who caused me to believe the vesting fund stored the VESTS, which caused me to wonder why the liquid STEEM in the vesting fund were backed by VESTS. I was just following what I thought people were saying.

I can't read people's minds. There needs to be clear documentation. The white paper afair doesn't explain all about VESTS and vesting fund terms (unless it is buried and I've forgotten?)

[alpesvillage]

Steem is crashing nowadays, it is worthless at all. Many retarded use this system to cheat free money, shame on them, especially bts founder dan

[iamnotback]

You can easily encode data in the amounts of payments. If amounts are hidden you can encode data in the number of outputs, or if number of outputs is fixed, you can encode data in the timing of payments.

There are a million ways. How contrived these measure get will be a function of how much effort is put into suppressing it.

Payments are numbers. Anyone can write any filter they want to reinterpret any data any where on the Internet as anything. Everything would be illegal. So obviously this will never stand up in any sane court-of-law.

You are being absurd.

Are you interested in improving things or just nonsense.

So if I'm reading smooth's comment correctly, couldn't multiple ui's interpret the information differently, allowing for decentralized use? I mean you could go after steem, but that (eventually) turns into a game of whack-o-mole.

(I understand things through words, so forgive me if I'm missing mathematical parts.)

My understanding of smooth's stance is that nefarious data (e.g. stolen launch codes for nuclear ICBMs) can be encoded any where in any digital datums.

I retorted that the validators of the blockchain will not allow an invalid ECC signature, and it is intractable to control the bits of a valid ECC signature.

He also countered by stating that the payment amount datums and even the number of inputs/outputs in a transaction datums could be encoded with data that has some other meaning. My retort is that the blockchain validators would not make such an interpretation, and thus any such interpretation must come from an external perspective (e.g. an external UI client or any external program). And I explained that we can make external interpretations of any data posted on the Internet, so basically everything every where could become illegal. Which is absurd. No court is going to rule that the blockchain is illegal because someone distributes a UI client that pulls launch codes out of blockchain payment amount datums spread across perhaps numerous disjoint transactions. They will instead declare that UI client illegal (since it contains a special decoding algorithm) and go after it.

If we allow unvalidated text and blob fields on the blockchain (i.e. not disjoint data spread across numerous transactions and intentionally not verified to be of a certain limited scope), then yes maybe the blockchain can be culpable. Perhaps smooth is stonewalling because apparently (?) Steem is allowing that.

I had offered what I think is a pretty good solution and perhaps (?) Steem doesn't do it the way I suggested.

I want to make the blockchain just a record of what happened (i.e. enforcing the consistency and consensus), not a store of the content. By making the two orthogonal, I posit we gain certain properties.

This is an example of the sort of design work I do.

Edit: smooth was correct that my original idea of using a hash was flawed (although he didn't reply to me and instead put it in his reply to MWD64), assuming the outputs of the hash can be so deterministic (which I think may be possible to filter out with a heuristic but it might be problematic). Normally it is impossible to produce a hash output that is for example comprehensible text. There is no way to find an input which will produce such an output. Thus a heuristic may work. But instead we can use ECC signatures to avoid the problems with a heuristic.

Forgive me if I feel somewhat skeptical sometimes as to the sincerity of others. I've taken people at their word in the past and it hasn't worked out always so well for me. I want to make money and do something good at the same time. I understand vested interests. If I am on my own project, I won't try to pretend I am unbiased any more. Well I think I am not pretending now. I think everyone knows I am going over the design of Steem with a fine-toothed comb and with the intent to find flaws in it. But it does me no good to invent flaws that aren't serious. Nitpicking will only cause me to fail. So I discuss here with intent to be sincere and objective as I can, often catching myself along the way and righting my ship when it is keeling too much to one side. It is an imperfect process.

[iamnotback]

Steem is crashing nowadays, it is worthless at all. Many retarded use this system to cheat free money, shame on them, especially bts founder dan

There was just a weekly power down that occurred which makes ~1% of the SP liquid. So perhaps this is generating selling demand.

Does anyone know where the powered down SP goes? Into STEEM or STEEM DOLLARS?

[iamnotback]

Sybils, if they exist, basically have to come from Steemit signups because the situation in the mining market with some high powered miners getting most of the blocks means that only a very small number of accounts can be created that way per day. My guess (without data) is that it is cheaper to cheat Steemit out of $7-10 plus an account name than to mine $7-10 of coins plus an account name.

Actually, your average i7 doing 30K hps can mine about 1 STEEM/day (more if it's water cooled and can max out).

And one account scammer can probably sign up dozens or hundreds of free accounts (with 3 STEEM each) per day.

I'd say perhaps thousands if they are really serious and have a botnet of IP addresses, because they can hypothetically still sell those Facebook accounts for more than it cost them to create them. Of course they have their time opportunity cost of capital, while they wait to sell them unless they are only creating them as buy orders are received.

Pretty much figure that eventually it can be that every sold fake Facebook account is first a Steem account before it is sold, if there exists some game theory strategy that is profitable using Sybil signups on Steem. I presume it is possible to change the private key and remove the ability of the Facebook login to access the account? But most who are buying FB accounts wouldn't realize to try to login with them on Steem at this juncture.

[Script3d]

thanks for informing op i was planning to invest
my btc to steem

[smooth]

Steem is crashing nowadays, it is worthless at all. Many retarded use this system to cheat free money, shame on them, especially bts founder dan

There was just a weekly power down that occurred which makes ~1% of the SP liquid. So perhaps this is generating selling demand.

Does anyone know where the powered down SP goes? Into STEEM or STEEM DOLLARS?

Powered down SP goes to STEEM.

If there was one huge power down (I haven't looked) that would probably be the 'steemit' account. All the rest of power downs happen at more or less random times and are not synchronized. I don't know what Steemit's plans are for their powered down STEEM coins other than the long term plan they have stated to give away 40% to new accounts and sell 20%. They have at various times turned on and off the power down on the 'steemit' account.

It certainly does seem plausible that traders would be watching that.

I think VESTS are only created or destroyed by power up, power down, or rewards (that pay in SP/VESTS), unless I'm forgetting something.

If you sell SD on an exchange, literally nothing changes in terms of supply, just the owner of the existing token. That seems kind of obvious, no?

Yes changes in the price of STEEM (as reported by the oracles i.e. witnesses) between the time SD is created and when it is destroyed will influence the money supply (virtual supply; converted to real supply if and when the SD->STEEM conversion takes place). If I'm not mistaken that is covered in the white paper.

There is no special tagging of VESTS nor the STEEM/SP in the vesting fund.

I think the confusion was I thought commentators were stating that no liquid STEEM were in the vesting fund and I thought only VESTS would go into a "vesting fund". So I was confused as to where the liquid STEEM were accounted.

Now I think I've been told the VESTS are not in the vesting fund, so therefor liquid STEEM are accounted for in it, and the SP are accounted for in the VESTS. Is this correct? If not please provide some complete definitions. This is  getting ridiculous. How can anyone know WTF is going on if the damn terms and design isn't documented. I try to guess based on different statements which makes it look as though it is my error, but it is a failure of clear communication.

I'm not sure I'm following this but let me try to explain it clearly.

The vesting fund contains STEEM POWER. That STEEM POWER can only be converted to STEEM (1:1) according to the power down rules (1% per week, etc.). STEEM can be converted to STEEM POWER (1:1) by powering up.

The vesting fund can be denominated as VESTS, which represent shares of the vesting fund. VESTS are created and destroyed by:

1. Power ups (creates new VESTS)
2. Power downs (destroys VESTS)
3. Rewards paid in SP (creates new VESTS, and SP is added to the vesting fund so as to not dilute the vesting fund)

The exchange rate is the number of SP in the vesting fund divided by the number of VESTS. The exchange rate is modified whenever new SP are added to the vesting fund as anti-diluation payments. This increases the exchange rate per VESTS but does not change the number of VESTS.

What I understand now is that when we "Convert" our SBD, it retires them and converts to liquid STEEM. Whereas, when we choose "Buy or Sell", then we exchange the SBD for liquid STEEM and the SBD is not destroyed.

Correct.