Bitcoin Forum
May 27, 2020, 10:25:42 AM *
News: Latest Bitcoin Core release: 0.19.1 [Torrent]
 
  Home Help Search Login Register More  
  Show Posts
Pages: « 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 [19] 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 ... 401 »
361  Other / Meta / Re: Cryptios on: May 08, 2019, 09:20:53 PM
The account was security-locked, not "banned for sending stupid merit". The user would see a message saying "Your account is locked because it is suspected to be compromised, email ...". Then if they email and provide some evidence that it's not compromised, it'd be unlocked.

Locking without a recovery email can be appropriate in cases similar to this where the user woke up, their access had suspicious characteristics, and they immediately started doing suspicious things. If their access characteristics change from one week to the next, without a long period of inactivity, then I'd be more likely to assume by default that they're using someone else's computer or they sold their account (which is allowed), and I'd rely on a complaint email to move from this default assumption.

Cryptios's main purpose is recovering accounts, but they can also do minor patroller-type things. They are neither admins nor global moderators. They have access to IP/security logs (not PMs), but each of their accesses to private info are logged and per-day-limited; even if their access was compromised, the damage would be limited. In the past 7 days, they viewed the basic info/logs of 461 users and viewed the IP logs of 51 users.
362  Other / Meta / Re: @theymos is it true that you forced OG to remove Laudas tagg ? on: May 06, 2019, 08:00:25 PM
I pressured several people to remove inappropriate, unconstructive ratings amongst each other. When OgNasty replied telling me that he perceived real trust issues with Lauda, I responded:
If this is what you believe, then you shouldn't remove the rating, but then it wouldn't be fair for me to pressure Lauda to remove ratings against you. Personally, I find Lauda's history gray, not red. And the trust system is only going to work if there's some level of forgiveness and de-escalation. But if you really believe that Lauda isn't safe to deal with, then you should leave the rating.

Then there was an exchange between OgNasty and Lauda in which Lauda kept a laudably Cool cool head, and OgNasty was able to reach a point where he could feel OK removing the rating.

The ratings did all end up being removed, which I'm happy with, and I appreciate the willingness to de-escalate and forgive from the people involved in this case. The fact that this issue came up at all indicates that the trust system isn't working perfectly (and I am considering future system changes), but it's still a good outcome.
363  Other / Meta / Re: This altcoin is trying to trade merits ? Will theymos allow this ? on: May 05, 2019, 03:08:56 AM
Are the stable coins are not backed by the fiat currencies?

They typically are (or claim to), but that's not really the point. Most people never intend to trade the stablecoin for the underlying currency. The point is for it to have ~zero volatility so that you can use it as a non-position when trading.
364  Other / Meta / Re: This altcoin is trying to trade merits ? Will theymos allow this ? on: May 04, 2019, 10:24:56 PM
If they're just using merit as a metric for an airdrop, without requiring that anyone send merit, then I don't have a problem with that. It's totally centralized, obviously, but it's not a terrible way to do it.

What I dont understand is how can he guarantee or promise that the price of the token will be $1. It is not really that easy is it?

This is totally unrelated to this particular token, which I don't know anything about and only spent 30 seconds skimming, but you could do a stablecoin in a mostly-decentralized way like this:
 - Call the stablecoin "SSCN". There's also another token which is part of the same system called "SBND".
 - When the price is below $1, the system automatically creates x SBND tokens and sells them for SSCN tokens in an automatic auction process. This destroys SSCN coins, reducing the supply and therefore increasing the price.
 - When the price is above $1, the system automatically buys y SBND tokens in return for SSCN tokens in an automatic auction process. This creates SSCN coins, increasing the supply and therefore reducing the price.

(Speculators would buy SBND tokens if they think that SSCN is going to become more popular, and sell them if they think that SSCN is going to become less popular.)

A decentralized system cannot know anything about prices, so some centralized entity needs to provide SSCN's price to the system as an input. This is an unavoidable point of centralization. But everything else, such as determining x and y above and conducting the auctions, could be done automatically.
365  Other / Meta / Re: Don't over-create Trust Appeals; keep minds/ emotions stable. No more Red on: May 04, 2019, 05:07:09 AM
We have so many threads like this nowadays, people twerking for merits.

There are a lot of pointless "summarize something obvious" posts, but IMO btcsmlcmnr's summary added something.

Forgiveness and de-escalation are key to getting Trust working smoothly:
 - Forgiveness: Often people make fairly small mistakes, but then they seemingly get red-trusted for life. This isn't really fair, and it discourages participation due to paranoia: if you think that you have a 1% chance of running afoul of some unwritten rule and getting red-trusted for life, you might just avoid the marketplace altogether. Red trust should mostly be based on an evaluation of what the person is likely to do in the future moreso than a punishment/mark-of-shame.
 - De-escalation: If some people end up locked in a feud where they're only really giving negative trust to each other in retaliation for negative trust, then one of them should propose burying the hatchet and removing the negative trust. Otherwise it never gets resolved, and everyone is worse-off for it.
366  Other / Meta / Re: DT update log on: May 03, 2019, 11:58:10 PM
Old:
Code:
theymos
HostFat
gmaxwell
TECSHARE
phantastisch
OgNasty
SebastianJu
malevolent
qwk
Vod
Anduck
mprep
Dabs
Foxpup
philipma1957
Cyrus
monkeynuts
Welsh
TMAN
Lauda
Mitchell
vizique
Ticked
yogg
dbshck
TheNewAnon135246
hybridsole
greenplastic
hilariousandco
EcuaMobi
Avirunes
Lesbian Cow
willi9974
cryptodevil
suchmoon
Rmcdermott927
achow101
owlcatz
nutildah
minerjones
sapta
tmfp
BitcoinPenny
yahoo62278
zazarb
bill gator
LFC_Bitcoin
ezeminer
MaoChao
bones261
LoyceV
actmyname
WhiteManWhite
The Pharmacist
LeGaulois
DarkStar_
TheFuzzStone
Jet Cash
bL4nkcode
Lafu
Gunthar
Hhampuz
xtraelv
Kryptowerk
krogothmanhattan
Halab
Veleor
theyoungmillionaire
o_e_l_e_o
chimk
iasenko
coinlocket$
asche
DdmrDdmr
Alex_Sr
Hellmouth42
taikuri13
Coolcryptovator
1miau
ICOEthics

New:
Code:
theymos
HostFat
gmaxwell
TECSHARE
phantastisch
OgNasty
SebastianJu
malevolent
qwk
Vod
Anduck
mprep
Dabs
Foxpup
philipma1957
Cyrus
monkeynuts
Welsh
TMAN
Lauda
Mitchell
vizique
Ticked
yogg
dbshck
TheNewAnon135246
hybridsole
greenplastic
hedgy73
hilariousandco
EcuaMobi
Avirunes
buckrogers
Lesbian Cow
willi9974
cryptodevil
suchmoon
achow101
teeGUMES
owlcatz
nutildah
JohnUser
dazedfool
minerjones
sapta
tmfp
BitcoinPenny
yahoo62278
zazarb
bill gator
LFC_Bitcoin
ezeminer
MaoChao
bones261
LoyceV
actmyname
WhiteManWhite
The Pharmacist
LeGaulois
DarkStar_
TheFuzzStone
Jet Cash
bL4nkcode
Lafu
polymerbit
Hhampuz
xtraelv
Kryptowerk
krogothmanhattan
roycilik
Halab
kenzawak
Veleor
theyoungmillionaire
o_e_l_e_o
chimk
iasenko
coinlocket$
asche
DdmrDdmr
anonymousminer
Alex_Sr
morvillz7z
taikuri13
Coolcryptovator
lovesmayfamilis
1miau
ICOEthics
367  Economy / Reputation / Re: Trust Sytem or trusted member list Must Be Removed? on: May 01, 2019, 11:39:17 PM
Note that olcaytu2005 is banned on his request.


There are aspects of the trust system that are clearly less than ideal, some of which are probably solvable (and some probably not). But caucusing with people who have very clear and serious questionable behavior isn't going to help you convince people of anything. Anduck had a good case for the trust system being broken. You (cryptohunter) might've had some case for the trust system being broken if you had handled your initial trust issues with rationality and patience instead of absolutely losing your mind and throwing lies around, and even now I could imagine your red trust eventually being cleared if you were able to attain a stable mindset. olcaytu2005 on the other hand hasn't even attempted to defend himself against the accusations (from several independent widely-trusted members, based on publicly-available data) of knowingly assisting scams, so he does not have a good case for the trust system being broken.
368  Other / Meta / Cryptios on: May 01, 2019, 10:55:35 PM
FYI: The humans handling the new account recoveries system which was implemented late last year are a company that Cyrus created called Cryptios. alanst, 3dOOm, and Rizzrack -- the "mysterious newbies" who some people noticed -- are three of their employees. They've been doing a great job.

Although Cyrus created the company with bitcointalk.org in mind as their first customer, they are accepting additional customers. Their main specialization right now is high-security, social-engineering-proof customer support & community moderation.
369  Bitcoin / Development & Technical Discussion / Re: Idea for uncensorable DNS on: April 29, 2019, 07:12:53 PM
Has not the idea of uncensorable DNS been proposed before, while Satoshi was around?

bitDNS - https://bitcointalk.org/index.php?topic=1790.0

I posted in that topic in 2010. Wink Still haven't given up on the idea.

The specific idea in that topic was basically Namecoin, but implemented sort of like a na´ve colored coins thing on Bitcoin. Eventually it became Namecoin. In any case, it has the censorability issues I mentioned in my OP.
370  Other / Politics & Society / Re: A radical left-right compromise US budget on: April 29, 2019, 07:04:29 PM
Damn, I guess I shouldn't enter the Democratic primary on this platform. And I was getting all ready for the debates...

If you tax $18 trillion worth of property at 10% and 14.3 trillion worth of sales at the same percentage rate

It's 10% property tax and 55% sales tax.

Everybody needs a place a place to live, or to do business.  The property owners are just going to pass on the costs to their tenants.  Regardless of whether you own the property, you'll be paying the tax.

Correct, which is why it's included in the real-UBI calculation. It's a wealth tax because someone just sitting around doing nothing on his property still has to pay tax, which is rather abhorrent to me, but it's still a lesser evil compared to income tax IMO.

The annual premiums for an unsubsidized health insurance plan is between $4700 and $12,800 per year (for a single person), depending on, primarily the deductible. At the low end, someone should expect to pay about a total of $5,000 in healthcare costs per year, per person. This would include the cost of health insurance and a small amount of out of pocket expenses. This amount would be for a generally healthy person.

Health care costs are too high due to over-regulation. That should be fixed separately.

The annual food stamp benefit is around $2,000 per person (based on a 3 person household).

The EITC is ~$3,500 with one child, or about $1,175 (rounded up to $2,000/year) per person assuming two parents and one child. It is ~$5,800 with two children, or $1,450/person with two parents. Budget assumptions need to be at least 1 child/person, otherwise the population would shrink, which would cause other budget problems.

The above three programs alone are valued at approximately $9,000 per year, and it is not uncommon for a working family to be eligible for all three of the above programs.

For someone that is not working, or only earning a very little amount, the additional government benefits will far exceed the UBI amount. Most states pay between $300 and $500/week in unemployment benefits. Assuming a $400/week benefit per working adult, or $200/week/person, (based on two working adults and two children) works out to $10,400/year. Someone on SSDI will receive ~$14,800 per year, or ~$7,400/person (based on two working adults and two children).

My UBI is per citizen, including children. So a family of four would get $49,960/year, blowing most existing welfare programs out of the water. I think that the only people who could get less would be some disabled/unhealthy people.

But if you want to increase it more, you can. Here are some example numbers for different UBI amounts:
Pre-tax-adjustment UBIPost-tax-adjustment UBITotal UBI costProperty taxSales tax
1249020608.506.3T10%55%
1749033580.8010.3T12%80%
250006425019.6T15%142%
35250167437.5051.2T25%350%

(Although over-100% salex tax might sound ridiculous, in practice today's taxes are probably above 100% if you took all of the payroll, corporate, income, gax, etc. tax and concentrated them into the point of sale.)

Quote
The problem with UBI when the majority of a population can work is those on the low end of the income spectrum will effectively be subsidizing those who are "rich". The purpose of welfare programs is to help the needy when they are most vulnerable, not to give money to everyone, regardless of need.

When rich people get the UBI, you can think of it as sort of progressive tax, like how even if you make $1 million in a year, you still pay 0% income tax on the first dollar.

For the sake of argument, and not based on any data, let's say that people own 35% of their yearly income in base-land-value (or this is represented in increased rent/other prices), and that they spend yearly amounts wildly-guessed below. Then you get:
IncomeSpending on goodsTotal taxEffective income tax rate
112000-14,008 -
10,00012,000-13,658-137%
20,00012,000-13,658-67%
30,00015,000-11,308-38%
40,00020,000-8,208-21%
50,00025,000-5,108-10%
75,00027,000-3,133-4%
100,00030,000-608-1%
250,00035,0007,3923%
500,00045,00021,6424%
1,000,000100,00069,3927%
1,500,000150,000114,3928%
10,000,000200,000439,3924%
50,000,000250,0001,866,8924%
100,000,000300,0003,644,3924%

You can see that it approximates a progressive income tax with a Friedman-style negative income tax. The progressiveness breaks down a bit at the truly high end (beyond what anyone seriously claims on their income taxes...), but this could perhaps be addressed by increasing the property tax overall and/or making it progressive at the very high end.

Quote
The biggest owner of land is the federal government, which cannot raise money by taxing itself. I haven't looked into your figures closely, but I suspect the claimed revenue from a property tax is overstated.

The Net Present Value of any asset is the current value (discounted value) of all future cash flows the asset will generate. If an asset will incur additional costs it did not previously incur, then its value will decline, all else being equal. Imposing a tax on real property will cause its value to decline, so you would need to either impose a higher tax rate, or budget for less revenue.

Government land isn't included. It's from that Federal Reserve document using figures from tables B.101, B.103, and B.104.

The average local property tax is 1.2%, and that includes improvements as well as base land value. IMO 10% wouldn't destroy the economy. (I'd actually originally wanted to do exclusively property tax, but it ended up not being enough revenue at reasonable tax rates, so I added sales tax. If it was like a 50% tax on total real-estate value, then I agree that land values would drop catastrophically.)

I didn't include land improvements because I think that assessments of this are rather invasive, and it also discourages economic development. If you do include improvements, then it's a total of $51.8 trillion.
371  Economy / Auctions / Advertise on this forum - Round 276 on: April 29, 2019, 03:00:49 PM
The forum sells ad space in the area beneath the first post of every topic page. This income is used primarily to cover hosting costs and to pay moderators for their work (there are many moderators, so each moderator gets only a small amount -- moderators should be seen as volunteers, not employees). Any leftover amount is typically either saved for future expenses or otherwise reinvested into the forum or the ecosystem.

Ads are allowed to contain any non-annoying HTML/CSS style. No images, JavaScript, or animation. Ads must appear 3 or fewer lines tall in my browser (Firefox, 900px wide). Ad text may not contain lies, misrepresentation, or inappropriate language. Ads may not link directly to any NSFW page. No ICOs[1], banks, funds, or anything else that a person can be said to "invest" in; I may very rarely make exceptions if you convince me that you are ultra legit, but don't count on it. Ads may be rejected for other reasons, and I may remove ads even after they are accepted.

There are 10 total ad slots which are randomly rotated. So one ad slot has a one in ten chance of appearing. Nine of the slots are for sale here. Ads appear only on topic pages with more than one post, and only for people using the default theme.

Duration

- Your ads are guaranteed to be up for at least 7 days.
- I usually try to keep ads up for no more than 8 or 9 days.
- Sometimes ads might be up for longer, but hopefully no longer than 12 days. Even if past rounds sometimes lasted for long periods of time, you should not rely on this for your ads.

Stats

Exact historical impression counts per slot:
https://bitcointalk.org/adrotate.php?adstats

Info about the current ad slots:
https://bitcointalk.org/adrotate.php?adinfo

Ad blocking

Hero/Legendary members, Donators, VIPs, and moderators have the ability to disable ads. I don't expect many people to use this option. These people don't increase the impression stats for your ads.

I try to bypass Adblock Plus filters as much as possible, though this is not guaranteed. It is difficult or impossible for ABP filters to block the ad space itself without blocking posts. However, filters can match against the URLs in your links, your CSS classes and style attributes, and the HTML structure of your ads.

To prevent matches against URLs: I have some JavaScript which fixes links blocked by ABP. You must tell me if you want this for your ads. When someone with ABP and JavaScript enabled views your ads, your links are changed to a special randomized bitcointalk.org URL which redirects to your site when visited. People without ABP are unaffected, even if they don't have JavaScript enabled. The downsides are:
- ABP users will see the redirection link when they hover over the link, even if they disable ABP for the forum.
- Getting referral stats might become even more difficult.
- Some users might get a warning when redirecting from https to http.

To prevent matching on CSS classes/styles: Don't use inline CSS. I can give your ad a CSS class that is randomized on each pageload, but you must request this.

To prevent matching against your HTML structure: Use only one <a> and no other tags if possible. If your ads get blocked because of matching done on something inside of your ad, you are responsible for noticing this and giving me new ad HTML.

Designing ads

Make sure that your ads look good when you download and edit this test page:
https://bitcointalk.org/ad_test.html
Also read the comments in that file.

Images are not allowed no matter how they are created (CSS, SVG, or data URI). Occasionally I will make an exception for small logos and such, but you must get pre-approval from me first.

The maximum size of any one ad is 51200 bytes.

I will send you more detailed styling rules if you win slots in this auction (or upon request).

Auction rules

You must be at least a Jr Member to bid. If you are not a Jr Member and you really want to bid, you should PM me first. Tell me in the PM what you're going to advertise. You might be required to pay some amount in advance. Everyone else: Please quickly PM newbies who try to bid here to warn them against impersonation scammers.

If you have never purchased forum ad space before, and it is not blatantly obvious what you're going to advertise, say what you're going to advertise in your first bid, or tell me in a PM.

Post your bids in this thread. Prices must be stated in BTC per slot. You must state the maximum number of slots you want. When the auction ends, the highest bidders will have their slots filled until all nine slots are filled.

So if someone bids for 9 slots @ 5 BTC and this is the highest bid, then he'll get all 9 slots. If the two highest bids are 9 slots @ 4 BTC and 1 slot @ 5 BTC, then the first person will get 8 slots and the second person will get 1 slot.

The notation "2 @ 5" means 2 slots for 5 BTC each. Not 2 slots for 5 BTC total.

- When you post a bid, the bids in your previous posts are considered to be automatically canceled. You can put multiple bids in one post, however.
- All bid prices must be evenly divisible by 0.02.
- The bidding starts at 0.02.
- I will end the auction at an arbitrary time. Unless I say otherwise, I typically try to end auctions within a few days of 10 days from the time of this post, but unexpected circumstances may sometimes force me to end the auction anytime between 4 and 22 days from the start. I have a small bias toward ending auctions on Fridays, Sundays, and Mondays.
- If two people bid at the same price, the person who bid first will have his slots filled first.
- Bids are considered invalid and will be ignored if they do not specify both a price and a max quantity, or if they could not possibly win any slots

If these rules are confusing, look at some of the past forum ad auctions to see how it's done.

I reserve the right to reject bids, even days after the bid is made.

Price flattening

At the end of the auction, after the winning bids are all determined, I will do a "price flattening" operation. This has no effect on which bids actually win. For each bid, in order of lowest to greatest price/slot, I will reduce each bid's price/slot to the highest value which is equal to or only the minimum increment greater than the next-lower bid. This allows you to bid higher prices without worrying so much, but you still mustn't bid more than you're willing to pay. Example:

Code:
This:
Slots  BTC/Slot  Person
    6      0.20       A
    1      0.16       B
    1      0.08       C
    1      0.08       D

Becomes:
Slots  BTC/Slot  Person
    6      0.12       A [step 4: reduced to 0.10+0.02=0.12]
    1      0.10       B [step 3: reduced to 0.08+0.02=0.10]
    1      0.08       C [step 2: same as the next-lowest, unchanged]
    1      0.08       D [step 1: the lowest bid is always unchanged]

Payment, etc.

You must pay for your slots within 24 hours of receiving the payment address. Otherwise your slots may be sold to someone else, and I might even give you a negative trust rating. I will send you the payment information via forum PM from this account ("theymos", user ID 35) after announcing the auction results in this thread. You might receive false payment information from scammers pretending to be me. They might even have somewhat similar usernames. Be careful.

[1]: For the purposes of forum ads, an ICO is any token, altcoin, or other altcoin-like thing which meets any of the following criteria: it is primarily run/backed by a company; it is substantially, fundamentally centralized in either operation or coin distribution; or it is not yet possible for two unprivileged users of the system to send coins directly to each other in a P2P way. The intention here is to allow community efforts to advertise things like Litecoin, but not to allow ICO funding, even when the ICO is disguised in various ways.
372  Economy / Auctions / Re: Advertise on this forum - Round 275 on: April 29, 2019, 02:56:18 PM
Auction ended, final result:
Slots BTC/Slot Person
2 0.08 acarli [flattened from 0.10]
4 0.06 lightlord
1 0.04 BoXXoB
2 0.04 SwC_Poker
373  Bitcoin / Development & Technical Discussion / Re: Idea for uncensorable DNS on: April 29, 2019, 05:55:49 AM
I think the tool you want is a locally decodable error correcting code. (some old discussion applying it to blocks)

Ah, good idea! It's not immediately obvious to me how you'd merge all of the data together, though.

Quote
I'm not sure why the timelock encryption couldn't be used to help with that.

In my boolean expression scheme, it seems necessary for miners to merge together all of the transactions with their true indexes, since this is important for performance and uncensorability, and merging can AFAICT only be done once. In a scheme based on error correcting codes, the time-delay encryption probably could be used here.

Quote
Depends on what you mean.

What I had in mind was an asymmetric lattice-based system where an encrypted database would be initialized by someone, but then the private key would be discarded. Then all network participants would have copies of the encrypted database and perform operations on it which would add/remove/update DNS data by encrypting the updates to the database's public key, but it (hopefully) wouldn't be possible to tweak the encrypted database directly in order to censor specific entries. Transactions would look like "apply UPDATE operation on your copy of the database using input <data encrypted to pubkey>".

However, as far as I can tell there'd be no way to get unencrypted data out of such a database in a controlled way, which you'd need to do in order to do lookups.
374  Other / Meta / Re: Will Theymos publish IP address of users? on: April 29, 2019, 05:28:50 AM
theymos will provide information to law enforcement without a warrant if he believes the person has committed a serious crime with an actual victim.

If theymos receives a valid subpoena that can be enforced in the US, he will have to give up IP address information per the order. If you are being accused of "promoting bitcoin" in your country, theymos will almost certainly fight any subpoena he receives  for your information, however he may or may not be successful. I also don't think a US court would enforce a subpoena if you were only "crime" was "promoting bitcoin"

Right. Though you shouldn't trust anyone with stuff like this; use Tor if you're concerned. In DTalk's case it's especially important, and simple proxies are likely insufficient, since HTTP/SOCKS5 proxies are unencrypted and a man-in-the-middle could apply eg. HTTPS stripping attacks.
375  Bitcoin / Development & Technical Discussion / Idea for uncensorable DNS on: April 29, 2019, 02:11:56 AM
Maybe someone already thought of this, but in case not:

Goal

Provide a decentralized system to convert memorable short names like "bitcointalk.bit" into a small amount of data (512 B, say), while making it impossible for anyone by any means to take down or block names so registered.

Naïve solution

You can just put the name data into a block chain like Namecoin does, but this isn't ideal for (at least) these two reasons:
 - Miners and network participants can blacklist certain names, preventing them from operating. In the case of miners, only a majority of mining power is necessary to ban a name network-wide.
 - Network participants could perhaps be legally forced not to provide-for-download transactions/data related to certain domains. For example, a node operator could receive a DMCA takedown. (The exact legal details are debatable, but it's within the realm of possibility.)

(BTW, decentralized database systems usually have the same problem. Participants in the Tor hidden service DHT could refuse to provide service to certain hidden services which they blacklist, and if this happened on a wide scale it'd bring down the hidden service. I sketched out a possible solution for that problem in this post.)

Enhancement 1: time-delay encryption

Assume name data is being mined into blocks. Instead of having the blocks be immediately readable, the name data portion can be temporarily encrypted in this way:
 1. Hash together the concatenated hashes of the blocks with depth 20 through 120 or so to get a seed.
 2. Using the seed, choose a random pubkey point on a secp256k1-like curve. You get the point directly, not by producing a private key.
 3. Once you solve the block, you will asymmetrically encrypt the namedata portion to that pubkey.
 4. In order to solve the block, you need to brute-force the last not-yet-decrypted block from eg. ~24 hours ago, and include the private key in your new block. This is unsuitable as a timestamping proof-of-work, so after you do this and encrypt the namedata, you also have to timestamp your block hash using OpenTimestamps or similar.

Unlike eg. RSA, AFAIK all valid-looking secp256k1 pubkeys have a private key. Difficulty should be auto-adjusted to maintain the ~24-hour delay by producing modified versions of secp256k1 with smaller integer sizes.

You could also use symmetric crypto with a random instead of seeded key, but then the miner knows the key, so you can't use it for PoW. Also, although I don't use it in this post, it might be useful in some cases to allow people to encrypt transactions/data to one of these pubkeys.

This prevents miners from soft-forking out blocks which contain banned names. By the time they can test whether the block contains a banned name, it'll be deep in the chain and therefore difficult to undo.

Enhancement 2: mixing together namedata

Each block needs to contain a function which takes a name and produces the output data for that name. Typically this is done as just a list, but this makes it easy to modify the list. If a name is banned, nodes could republish block data with that name's data redacted. My solution to this is to construct the function so that all of the name data is jumbled together, it's not immediately obvious which names are contained in it (though you can test individual names), and you can't remove one name without affecting everything else. The best thing I've come up with for constructing such a function is:

 1. Let hash(name) = encHash and hash(encHash) = idxHash. The name's data is encrypted with the key encHash before being published to the network.
 2. Each miner generates a random key and encrypts idxHash and the name data before dealing with them, purely to ensure that they are uniformly random. This key is published in the block's unencrypted segment.
 3. Looking at all idxHash values in his candidate block, the miner attempts to find a most-minimal subset of their bits such that each idxHash is uniquely identified by as few bits as possible. (Note that duplicate idxHash values are not allowed.) The chosen bits can be in any order, and need not be contiguous. For example, a miner might find that bits 2,3,10,15,22,1,40,32,100,125 results in unique IDs for 500 entries. This is published in the final block.
 4. Using the smaller index value for each name, the miner constructs a truth table with the index as input bits and the double-encrypted name data as the output bits. When the block is nearly complete, the miner converts this truth table into an as-minimal-as-possible set of boolean formulas.

Example:
Say that you have these keys (name hashes) and values:
Code:
KEY     VALUE
6A80AE9 A1
CDE2E9D 22
14DB3CF B3
1F43947 24
2CA8297 45

(In reality the keys would be 256 bits long, the values would be perhaps 4096 bits long, and there would be many more pairs.)

First, I choose a small subset of bits from the input such that each pair has a unique value. I choose bits 11-8, giving a binary truth table like this:
Code:
IN   OUT
1010 10100001
1110 00100010
0011 10110011
1001 00100100
0010 01000101

And then I create boolean formulas from that, dictating the output of each bit 7-0:
Code:
e1 = i3&!i2&!i0;
e2 = i1&i0;
e3 = !i3&!i0;

o7 = e1 | e2;
o6 = e3;
o5 = !i1 | e1 | e2 | i2;
o4 = e2;
o3 = 0;
o2 = !i1 | e3;
o1 = e2 | i2;
o0 = e1 | e2 | e3;

That's the function. When given the a key, it produces the correct output. Eg. for input 0011 you get:
Code:
i3 = 0
i2 = 0
i1 = 1
i0 = 1

e1 = i3&!i2&!i0 = 0;
e2 = i1&i0 = 1;
e3 = !i3&!i0 = 0;

o7 = e1 | e2 = 1;
o6 = e3 = 0;
o5 = !i1 | e1 | e2 | i2 = 1;
o4 = e2 = 1;
o3 = 0;
o2 = !i1 | e3 = 0;
o1 = e2 | i2 = 1;
o0 = e1 | e2 | e3 = 1;

But you can't tell by looking at the function which keys were used to create it. You can test that it contains a name by trying to resolve the name using the function's data and seeing if it works, but if you want to remove it from the function, there's no way to do so without possibly ruining other keys, since you don't know what other keys are represented in the function, and everything is tangled together.

Sketch of the rest of a DNS

Every block should have a bloom filter matching the idxHash values within it, in the encrypted segment. To resolve a name, you'd go through every block in the past and check the name's idxHash against the bloom filter. If it passes the bloom filter, you reconstruct the data using the function explained above and try to decrypt it using its encHash. If it decrypts with a good MAC, then it's OK. You get the current name data by replaying its entire history of valid namedata updates, the first-ever of which will have a pubkey with which the rest are signed. Names might expire if not updated for a few years, to allow lost ones to return to the pool.

For payment, this is the best I could come up with: To register/update a name, you'd pay a "registrar" and give them the encrypted namedata update & idxHash. The registrar would send the namedata update to all miners along with a willing-to-pay price, signed with the registrar's own key. When a miner includes it (which the registrar & other miners can check by trying to find the idxHash in every future block once it's decrypted), the registrar pays a miner-address listed in the block the promised amount. Miners keep track of which registrars are actually paying them the promised amounts, and will blacklist any who stiff them. To start a new registrar, you might pay a miner an initial fee to get them to start listening to your transactions.

Miners would be constantly trying to brute-force the last unbroken encryption key from about a day ago. Unlike with normal cryptocurrencies, they wouldn't actually construct their final block until they found the solution. When they find it, they'd use OpenTimestamps to timestamp their block, wait for a timestamp-on-Bitcoin confirmation, and then broadcast it. To prevent withholding attacks, nodes would refuse to ever reorg unless the two candidate blocks are actually seen within minutes of each other. The brute-force difficulty might adjust only if the block interval goes outside of a pretty large range, which would allow for block generation to be more responsive to demand.

Flaws

When you first register a name, nobody has any idea what it is since registrars and miners only get the hash and the encrypted data. But once the name becomes publicly-known, registrars and miners could refuse to process matching idxHash updates. This may be somewhat annoying, but since a softfork banning the name network-wide isn't possible, enough darknet registrars/miners should exist to make this not too much of a problem.

To resolve a name, you have to download and scan through every past block. I think that blocks should be pretty small (perhaps about 25-50kB per 10k domain updates?), and the max block size could be set very restrictively. Scanning a single block should be extremely fast, and it's also parallelizable. But after many years, the history could accumulate to the point of being problematic, and I can't see a perfect way of resolving this. One optimization would be to require that every domain send at least one update a year, and require that every update contain a pointer to the first-ever update which created the name; then, if there are no conflicts in a year, you can just trust that one, and if there are conflicts, you can check through a subset of the whole history to figure out which claim is correct.

To prevent miners from filling up every block with domain-squatted domains, maybe there should be some additional cost for every registration. One way to do this would be to require a hashcash proof-of-work in the data for every new domain -- if it's not there, then the registration is ignored at resolve-time as if it didn't exist.

I wonder if it's theoretically possible to create a homomorphic encryption scheme where an encrypted registry database is maintained which nobody has the decryption keys for, but it's still possible to do lookups. That'd be ideal.
376  Other / Politics & Society / A radical left-right compromise US budget on: April 27, 2019, 03:02:36 AM
Although I'm an anarcho-capitalist and in an ideal world I'd like there to be no government at all, I'm very interested in government/politics, and I don't scoff at gradual changes in the right direction. For a while I've been thinking that a left-right compromise is possible by increasing total tax collected while simultaneously reducing intrusiveness, regulation, and size of government. Here is a specific plan:

1 - Remove income tax and replace it with a national sales tax and a national property tax

Income tax is bad for the economy -- when you tax something you always get less of it --, and it's also highly invasive because almost everything you do might generate some income.

Property tax is a sort of wealth tax, which is quite anti-freedom, but it's also a pretty non-invasive tax since the government just has to go to the land and demand that the tax be handed over. They don't even need to know who lives on the land. It's also impossible to evade, which would be relevant if the entire world switched to private cryptocurrencies. Sales tax is also pretty non-invasive. Only businesses have to directly deal with it, and they just have to count up their sales and pay the appropriate tax.

The total land value of the US (not including improvements) is about $18 trillion.1 US consumers buy about $14.3 trilion in goods per year.2

I propose a 10% annual national property tax (on the value of the land alone, not including improvements, unlike most local property taxes) and a 55% national sales tax. This gives a (naïvely estimated) revenue of $9.7 trillion, much higher than the current US tax revenue of about $3.4 trillion.

A 55% sales tax sounds high, but keep in mind that the US currently has a 25% "sales tax" on imported steel, an 8% sales tax on diesel, and roughly a 22% sales tax on cigarettes. People can live with it, especially if their income tax is being eliminated.

2 - universal basic income

In order to offset the regressive nature of the sales tax and also act as a sort of very easy-to-administer welfare system, I propose a UBI of $20,608 per US citizen per year. This is calculated from the current federal poverty guideline of $12,490 plus 55% to cover the sales tax and 10% to cover the property tax. No means testing, since that complexifies the system -- think of it as undoing some of the property/sales tax in a progressive way.

More humans are good. A new human is someone who's almost certainly going to produce more in his lifetime than he's going to consume; it's not "just another mouth to feed". So in my plan I decided to include even children in the UBI, which massively subsidizes having children. You could instead choose to not include the 20% of US people under 16, or to give them less per year, which would reduce costs.

The US population is about 306 million3 when you subtract the 7% of the US population who are non-citizens 4. So the UBI has a total cost of $6.3 trillion per year, far higher than the current welfare spending.

3 - eliminate all other welfare

Eliminate social security, medicare, medicaid, and all other welfare. The UBI covers it instead. This saves about $3.3 trillion from the budget.5

Most people should be working. So don't complain about the (real) $12k/year not being enough for every little thing. For the miniscule percentage of the population who can't work at all, they could move to somewhere far away from major cities, where $12k/year is often enough to live on, or they could live with one or more roommates to reduce costs. If someone needs major surgery or something and they can't afford it, they should look to charities and their neighbors for help.

4 - eliminate all work-related regulations

Since nobody has to work in a particular job in order to survive, there's no argument for any work-related regulations. All work is now totally optional. So eliminate minimum wage laws, regulations related to hours worked, child labor laws, etc. This will boost GDP and probably increase tax revenue, possibly allowing for lower tax rates than given above, though I'm not qualified to estimate this.

Summary

So the budget is:
 - Property tax revenue: $1.8 trillion
 - Sales tax revenue: $7.9 trillion
 - Total tax revenue: $9.7 trillion
 - Non-welfare spending retained*: $3.3 trillion5
 - UBI: $6.3 trillion
 - Total cost: $9.6 trillion

(*Note that I would want to massively reduce war spending, but that's out of scope for this discussion.)

Although tax revenue is massively increased and total welfare is about doubled, the size of government in its intrusiveness and number employees massively shrinks. This IMO would be a pretty good outcome, and perhaps it's not politically impossible that the "far-left" and "far-right" could join forces in order to achieve something like this.


[1] https://www.federalreserve.gov/releases/z1/Current/z1.pdf
[2] https://apps.bea.gov/iTable/iTable.cfm?reqid=19&step=2#reqid=19&step=2&isuri=1&1921=survey
[3] https://www.kff.org/other/state-indicator/distribution-by-citizenship-status/?currentTimeframe=0&sortModel=%7B%22colId%22:%22Location%22,%22sort%22:%22asc%22%7D
[4] https://www.census.gov/popclock/
[5] https://www.usaspending.gov/#/explorer/budget_function
377  Other / Meta / Re: Viewing TRUST when not logged in on: April 26, 2019, 10:21:11 PM
Logged-out users will now see a warning in trust-enabled sections if more DT members neg-trust the topic starter than positive-trust him.

This increases the responsibility of DT members not to give negative trust for stupid reasons, but only for things that cause you to believe that the person is a scammer.
378  Economy / Speculation / Re: Wall Observer BTC/USD - Bitcoin price movement tracking & discussion on: April 26, 2019, 03:03:03 AM
BitFinx responds: https://www.bitfinex.com/posts/356

Confirms that a ton of money is at least "tied up".
379  Other / Meta / Re: [Another one, if possible, please] Is it possible me to find my "lucky" number on: April 25, 2019, 09:40:43 PM

914


You're not there yet.  Tongue


775, the minimum. Only a 0.4% chance of that.
380  Other / Meta / Re: Retention/privacy info on: April 25, 2019, 07:51:31 PM
Is this change in response to some government order? Just asking...

No. Previously IPs were logged sporadically but usually kept indefinitely; this is an overall significant reduction in retention.

It's a bad idea to provide any IP log to the user themselves. Compromised accounts happen and the situation could become worse if the attacker can access your IP logs.

Right. Though if someone really wants to know, I might consider manually giving them their logs after verifying that their account doesn't look hacked.
Pages: « 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 [19] 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 ... 401 »
Sponsored by , a Bitcoin-accepting VPN.
Powered by MySQL Powered by PHP Powered by SMF 1.1.19 | SMF © 2006-2009, Simple Machines Valid XHTML 1.0! Valid CSS!