Dandelion FAQ
Dandelion FAQ
Dandelion FAQ
What is Dandelion?
At it's core, the Dandelion Network is as an autonomous, independent entity that operates as a
universal public utility for trustable computing. The network provides trustable computing as a
commodity in the same way that a decentralized electricity grid supplies secure, resilient power
as a commodity.
Dandelion (the company) is a small startup based in Toronto, and dedicated to bringing the
Dandelion vision to the world.
A world in which all 8 billion people have access to secure, resilient, trustable transaction
services, regardless of geography or economic status.
Dandelion will disrupt existing transaction networks in order to capture, enable global financial
inclusion.
Dandelions are humble, beautiful, and unstoppable. We could not think of a better symbol for
our vision, our company, and our people.
Dandelion is built on the heritage of open transaction systems, pioneered by Bitcoin and open
computation, pioneered by Ethereum. These systems changed the world, but face serious
sustainability and effiency problems. Paul Chafe and five of his students from George Brown
College set out to find a better way. In collaboration with Atefeh Mashatan, director of the
Cybersecurity Research Lab at Toronto Metropolitan University, they won a peer reviewed
research grant from the Mathematics, Information Technology, and Complex Systems (MITACS)
Centre of Excellence, part of the Canadian government's National Centres of Excellence
initiative. Alex Munro, a veteran of BNP Paribas, left a successful VC career to lead the team on
the business and financial side. Today the Dandelion team has twenty core members on three
continents, but we are at heart a Toronto company, built around the students who came to do
research and stayed to built the future. Our technical team includes three PhD in electrical
engineering, computer science and mathematics, and our business advisory board contains
several successful CEO. We draft talent from the George Brown blockchain program, the
Cybersecurity Research Lab at TMU, the University of Waterloo's computer engineering
program, and the Computer Science Innovation Lab at the University of Toronto.
We screen for the ability to learn fast, entreprenurial mindset, and strong leadership potential.
People with this trilogy will learn what they have to learn, do what they have to do, to get done
what they have to get done. More important, they will have a great time doing it, and inspire
others to come do it too. We built our culture before we built our product, and we love them
both.
To date, Dandelion's research and development has been funded with research grants issued by
the nonprofit research organization Mathematics of Information Technology and Complex
Systems (MITACS), which is in turn supported by the Government of Canada. This research is
conducted through, and with the support of the blockchain program at George Brown College,
and the Cybersecurity Research Lab at Toronto Metropolitan University. Other company
operations have been funded by small investments from Revtech Lab Fintech Accellerator, from
our founders, and from private individuals. Dandelion has not accepted venture capital funding
at this time.
As systems have grown more complex and connected, we have come to rely on them more and
more. Today’s world is completely connected, but the systems we rely on are dangerously
vulnerable to failure and even digital attack, at every level from the silicon up to the datacentre.
Hacks and crashes cost us over a trillion dollars annually – and have cost lives as well. Open
computation networks with the property of decentralized trust eliminate these vulnerabilities.
Anyone can deploy and run code which is both fully transparent and is guaranteed to execute as-
written and fairly. Coders still have to write their code correctly, but no-one can exploit the
underlying system to make it run incorrectly.
Anything! To date decentralized computation has been used mainly for transactions, as
pioneered by Bitcoin. Transactions are just a specific form of compuation, and they were the go-
to application for early work in the field. Ethereum generalized this idea to the include any
computation, using transactions as inputs and outputs. Simple smart-contracts are the primary
use-case on Ethereum and similar platforms, because their costs are so high. More complex
applications can run in theory, but in practice are prohibitively expensive.
Dandelion still does transactions (at tap-and-pay speed and global scale) – in fact a transaction is
the basic unit of both value transmission and data transmission on the network, the necessary
first step in any trustable computation. However we make trustable computation so efficient and
so accessible that transactions are just the starting point. Dandelion can run secure messaging
without a central server, operate social media networks that can’t harvest user data, run markets
that cannot be front run or high-frequency-traded, support distributed wireless networks (so you
can fire your phone company), provide artist-driven content delivery without middlemen, and
much more. As with the early internet, the applications weren’t even imaginable when the
technology first launched.
Really fast! On a high speed 5G connection you can complete a tap-and-pay transaction in under
a second. The network as a whole can handle over a quarter million of those transactions every
second, on ordinary desktop servers with good bandwidth. More important though, Dandelion
scales naturally with demand. This is an important distinction from systems like Bitcoin, which
cannot scale natively and so must use “Layer 2” solutions that inevitably compromise security in
search of speed. In fact Dandelion is faster on Layer 1 than Bitcoin Lightning, Ethereum
Plasma, and even Polgygon are on layer 2.
Really efficient! In fact, we believe Dandelion is the most efficient decentralized network that
can exist. Like all open decentralized systems, Dandelion is Asynchronous Byzantine Fault
Tolerant (ABFT). This means the network will work properly as long as a majority of its nodes
act honestly. However, fault tolerance comes at a cost. In a pure-peer network, the number of
messages needed to come to consensus rises proportional to the square of the number of nodes,
because every node has to update every other node at every consensus cycle. This becomes
untenable long before there are enough nodes to consider the network effectively decentralized.
Systems built on Proof-of-Work and Proof-of-Stake use a single leader to decide on consensus,
after which every other node simply has to agree that the decision is correct. These reduces the
co-ordination cost to simply proportional to the number of nodes, but comes with another cost.
These systems deliverately waste vast amounts of electrical or economic power in the race to win
the leader election. This makes them slow, and expensive (also environmentally and
economically unsustainable.
On Dandelion, the network's clients responsible for leading their own transactions, and handling
all the messages necessary for consensus. Dandelion nodes don't have to message to each other
at all to complete a simple transaction (there is some communication necessary for network-level
housekeeping). They are then free to devote all their time and bandwidth to servicing clients.
Transactions and smart contracts (which have transactions as inputs and outputs) need to run
with the full decentralized security of the network. Anything less simply isn't safe (it should be
noted, Layer 2 system accept this risk, this is why they're a bad idea). However, there are lots of
applications that need some level of redundancy, but not quite the full security of the network.
The Decentralized Data Centre provides this - computation, storage and bandwidth, available on
demand, to do whatever you want with. Imagine AWS or Microsoft Azure - but with a large
corporation telling you what you can or cannot run.
USING DANDELION
The best wallet is our Breeze App, which integrates tap-and-pay and pay-a-contact with through-
chain secure secret messaging. However we are integrating with other wallets, with browser and
hardware platforms as the priority
Yes! The Dandelion network is ideally suited for online microtransactions, due its high speed,
low cost, and fast finalization time.
Yes! Our Dandelion Breeze App runs on your phone, and does Tap-and-Pay over-the-counter,
and out-of-the-box.
Yes! With Breeze you can use simple QR-codes to request, initiate, and complete any payment.
Dandelion coins give access to units of trustable computation. Even if your counterparty don’t
need this service, lots of people do. This gives them the confidence that they can accept
Dandelion, because they know they can trade it away later. This actually works the same way
gold and silver work.
SECURITY
Yes! In concurrent systems theory, a system which is live will always be able to make
computational progress. It has neither deadlocks nor livelocks. In other words, valid
transactions will always be able to complete. To prove this, we built a mathematical model of
Dandelion using TLA+, a formal math system built by Leslie Lamport, one of the pioneers of the
field. We ran this model for two weeks on the University of Toronto's Niagara supercomputer,
visiting every state network could enter (over a trillion in total). This conclusively demonstrated
that the system has no deadlock or livelock states. Valid transactions can always complete given
a majority of honest nodes, so long as the clients do the work to complete them. As a bonus - we
got to use a supercomputer!
Yes! In concurrent systems theory, a system which is safe will not allow unintended states to
happen. In Dandelion's case, we want the system to by Asynchronous, Byzantine, Fault Tolerate
(ABFT), which means that the safety property must hold even if messages can be lost or
indefinitely delayed, nodes can drop on and off the network at random, and even if nodes behave
maliciously, lying about their own state or the state of other nodes. To prove this, we built a
provable security model of Dandelion. This is a mathematical analysis technique in which we
formalized the capabilities of an adversary who might be trying to break the system. This
analysis proved that invalid transactions will always fail, given a majority of honest nodes, and
regardless of what a malicious adversary does or how powerful it is. This was published as the
lead article the 2022 Cybersecurity issue of the peer-reviewed journal Mathematics. We're very
proud of it.
No. A Sybil-attack happens when an attacker is able to inexpensively add enough nodes to the
network to overwhelm consensus. Dandelion consensus is designed to decentralize, and does not
allow Sybil attack.
Distributed key recovery is a way of making your private key disaster-resilient, by splitting it
into several parts and storing those parts in widely separated locations (A bank vault, with trusted
friends, with your lawyer, etc). The neat part is, you can configure the split key so that you don't
need all the parts to recover your private key. For example, with ten parts, you could use any
five to recover them.
Today, cryptocurrency addresses are long strings of hexadecimal. They are hard to read, and
easy to make mistakes with, which makes them user unfriendly. Dandelion's Breeze app uses a
special algorithm to turn a Dandelion wallet address into a unique, visual symbol, kind of like a
digital flower. The algorithm is set up so that a single change to an address produces a radically
different symbol. The human eye is very good at picking up visual differences like this, which
makes mistakes immediately visible. This kind of user friendly security is built into everything
we do, and is essential if peer-to-peer transactions are to become the default mode of value
transfer for everyday use.
TECHNOLOGY
In a simple peer-to-peer network, every node has to update every other node on every step of
consensus. The number of messages required rises as the square of the number of nodes
(O(n^2), and quickly reaches an unworkable volume, no matter how much bandwidth you throw
at the problem. Networks like Bitcoin and Ethereum improve on this by electing a leader (either
through work done or stake committed). All communications go through the leader, which is
rotated block by block. This reduces the communications burden to simply proportional to the
number of nodes (O(n)) which is better. However these proofs wastes a lot of electrical or
economic power, forces the unscalable, serial blockchain structure, and leads to vulnerabilities
like in-block front-running and empty block mining.
Dandelion uses the fully parallel blocklattice structure, asynchronous clear and settle, and the
client-leader consensus architecture. Dandelion's communication burden is flat (O(1)),
regardless of the number of nodes on the network . More important, Dandelion scales with
demand, with a built-in upgrade incentive for nodes. The network will always be driven towards
the point where computational cost meets demand. In fact, we believe Dandelion is the most
efficient decentralized network that can exist.
What is parallelism, and why is it important?
Parallel systems can do many things at once, where serial systems can only do one thing at a
time. Conventional blockchains are inherently serial structures – literally a chain of
cryptographically linked blocks, built one link after another. Dandelion uses a blocklattice, a
fabric of many threads, all advancing simultaneously. This eliminates the bottleneck that makes
blockchain slow and expensive, without the security compromises that have seen so many “layer
two” and similar solutions get hacked.
Synchronous systems move in lockstep, like an assembly line. If a single component stops, the
entire system has to wait until it catches up– a situation known as blocking in computer science.
Asynchronous systems are like a swarm of bees. The hive as a whole is co-ordinated, but each
bee works on its own. Dandelion does this by splitting each transaction into two parts. A
transaction is cleared when a sending client commits funds to a specific receiver, and then settled
when the receiving client orders those funds in its own account (note the specific order doesn't
matter, as all receipts are additions. Because these operations are separated, the network is non-
blocking.
For a simple consensus network, every node must agree on every transaction – this is what
creates decentralized trust. The drawback is that every node has to talk to every other node to
reach consensus, and the number of connections required to reach consensus rises proportional to
the square of the number of nodes. To get around this, networks like Bitcoin and Ethereum use a
leader-election system. One node is elected to decide what transactions should be processed, and
then tells the rest of the nodes. These nodes simply verify that what the leader decided was
correct. Proof-of-Work and Proof-of-Stake are simply leader election systems that incentivize
the leader to decide -mostly-honestly.
This solves reduces the communications burdern to merely linear with the number of nodes, but
creates two more issues. The leader is in a privileged position, and its decisions can favour itself.
This shows up in practices like empty block mining and miner-extracted value, both now
common features of conventional blockchain mining, This reap extra rewards for the leader node
at the expense of the network as a whole. The larger problem is that the leader is itself a
bottleneck, which strictly limits the number of transactions which can be processed.
Client-Leader solves this problem by recognizing that clients have an inherent interest in
completing their own transactions. Each client is the leader of its own thread of the blocklattice,
clearing and settling transactions (asynchronously) as required. On Dandelion, the nodes don’t
talk to each other at all to handle a simple transaction. The co-ordination and rebroadcast
function is handled entirely by the client. The bandwidth burden at the node is constant no
matter how many nodes are on the network.
The parallel, asynchronous, client led architechture lets Dandelion scale in three ways. The
blocklattice scales just by adding more threads The network scales by adding blocklattices as
required. Most important, Dandelion scales with demand. A client needs a 2/3+1 consensus to
advance a transaction, and the slowest 1/3-1 of nodes run will get left of of consensus - and
consensus rewards - if they aren't fast enough to service the client-determined time constraint.
As demand grows, nodes are incented to upgrade their capacity to keep up. No other network
can scale like this.
It is possible to trust unfair systems, as long as they are consistently unfair. Stock exchanges
work, even though high frequency traders, brokers, and other privileged players use their
privilege to siphon value from simple customers. Blockchains work despite miner extracted
value, in-block frontrunning and other leader-exploitation techniques open to miners but not
users. However this creates market inefficiency, because otherwise profitable trades do not get
made when their marginal value doesn’t exceed the value that may be siphoned. Dandelion smart
contracts use Fair Ordering, which guarantees that no transaction can be prioritized over another
in smart contract execution.
Yes! Dandelion can easily operate any number of shards. This enables scaling beyond demand
scaling
Dandelion’s source code will be released on main-net launch. Anyone will be able to download
it, run a node, and modify it as they like, for free, and forever.
ECONOMICS
The Dandelion Network provides secure, resilient computation as a commodity resource to the
world, just as a distributed electrical grid provides electricity. Dandelion tokens let you use this
resource, the same way bus tokens let you ride the bus. New tokens are minted by nodes, who
are rewarded for participating in consensus, and burned when they are used to complete
transactions or computational tasks (for eg, executing smart contracts) This is quite different
from typical, fixed cap systems found in conventional blockchains. The dynamics are designed
to encourage adoption, encourage use, while growing the supply to a stable level, avoiding the
instability that plagues existing systems. The idea is to maximize the rate of network value
growth, rather than the product of token volume and token value.
Yes, but not the way it's usually implemented. Proof of Work works by encapsulating energy in a
verifiable token – This is what made gold the literal gold-standard for over six-thousand years –
gold was the most efficient verifiable energy encapsulator available to the ancient world.
However, where conventional blockchain Proof of Work presents nodes with an open-ended race
to waste electrical power on hashing (an unsustainable model in the modern world), Dandelion
uses the completion of computational tasks (including transactions) for users. This work directly
provides a useful service, and it uses only the energy required to complete the computation. As a
result Dandelion is orders of magnitude more efficient than Bitcoin and similar systems. There
is more to it than simple computation, but that's it in a nutshell.
Yes. Dandelion nodes offer computation and transaction services to clients, and compete for the
opportunity to offer those services. The competition is structured much like a gold-rush. At the
beginning there are a small number of high quality sites. As the gold rush goes on (or the
network evolves), the quality declines, but the number of sites available goes up. This fairly
balances risk and payoff, rewarding early adopters while still providing incentive for people to
join later on - and steadily driving decentralization as the rush continues.
As with all things Dandelion, the decentralization schedule is dynamic. As demand goes up,
more nodes will be able to join the network. The best thing to do to help decentralize is to
generate demand. As a very rough rule of thumb, node count will double every year. However,
remember that later nodes will require more effort for the same reward, reflecting the lower risk
they must accept in joining an already-established network.
Economics are only inflationary or deflationary in a given context. Dandelion’s context is the
global demand for secure, resilient computation, which is growing rapidly. Within that context,
Dandelion's mint/burn system is aimed at maximizing the growth of network value rather than
the product of token value and token supply. This shift in emphasis is what will allow us to
rapidly pass established systems in value provided.
Sending clients pay a small fee for each transaction, or more broadly, each computational task.
This fee is burned. Simultaneously, the nodes which participate in consensus are paid a reward
in new-minted coins. Note that the fee paid and the reward granted are not necessarily the same
amount. This allows the network to automatically adjust the total coin volume to demand. More
demand reduces coin flow, increasing the external coin price, and triggering the scaling
incentive. As network capacity scales up, coin demand falls in relation to it. More coins are
produced, and the the external price optimizes to maximize use (and so value provided) at the
new, higher capacity. The idea is to maximize the rate of increase of value provided by the
network, rather than simply the product of coin price and coin volume.
Coins are introduced into the system on a exponentially declining curve that begins with
(nominally) ten million generated in the first year, declining by 20% per year to an approximate
of fifty million in thirty years. Note that this is the the baseline rate given average network
demand as measured against supply at any given moment. In detail, Dandelion uses a decoupled
mint/burn system that dynamically adjusts supply flow to balance demand, removing volatility.
The idea is to ensure the marginal value of a single coin to be a constant function of risk, reward,
and utility, in order to reward early adopters while continuing to incenticive adoption and use
along the entire curve.
What is the demand tarrif?
Dandelion dynamically allocates sets the price for transactions and computations based on a
number of factors. As demand increases towards total network capacity, the cost of transactions
and computation increases non -linearly. Under normal load, this has no significant impact on
operation. Under natural demand spikes, prices rise and fall smoothly. However, if a malicious
attacker attempts a transaction flooding attack, it quickly becomes ruinously expensive for them,
while having minimal impact on normal users. For example, if the network saturates at a quarter
million transactions per second, and normal demand is half that, then normal transactions will
cost a very small fraction of a penny. However, if an attacker floods the network, transaction
fees will rise rapidly. If they are $0.10 at 90% capacity and a $1.00 at 99% capacity, the attacker,
must pay somewhere between $12,500 and $125,000 per second to sustain the attack!
Meanwhile, normal users can easily afford this fee increase for the few transactions they will
complete until the attacker runs out of funds - or they can simply wait until the attack is over.
This is perhaps the only cyberattack a network could welcome!
LAUNCH
Dandelion will stage deployment through testnet and mainnet phases. Both testnet coins and
nodes will transition to mainnet after we have proven network stability.
As with Ethereum Goerli, testnet nodes will operate live, and testnet coins will have value (after
all, they give access to an operating service). They will initially be available to developers and
others through a faucet and OTC. We are talking to several exchanges now for early listing, with
one exchange confirmed.
Yes! We will be holding an open auction for testnet node positions. Anyone can join the
auction, and we are working to make sure it is widely available to the crypto community.
Yes, on mainnet all nodes will have to compete to participate in consensus, including all testnet
nodes. Testnet operators gain skill and experience in the competition, at the cost of up-front risk
and effort. However winning a place in consensus requires competing on a level field, and the
decentralization mechanism is always opening up new opportunities for new entrants.
Why would I buy a testnet node at auction if I can get a node for free on mainnet launch?
Remember the Dandelion proof protocol works like a gold rush. Earlier arrivals have an easier
time finding a site, and sites found later will take more effort to earn the same reward (trading off
the lower risk of waiting before investing). On mainnet launch, anyone can download and run
the software - and we hope a lot of people do. However, not everyone (in fact, not most people)
will be to win a place in consensus, and the reward rate is lower the later you join. As with the
coin generation schedule, the intent is to provide a constant function of risk, reward, and utility.
This is what makes it worth investing in testnet participation.
Late 2023.
We are targeting four months, but will run as long as necessary to resolve any issues.
This depends on how the auction goes, but in all cases Dandelion will retain a sub-consensus
fraction of nodes. This is necessary for the decentralized trust property to hold.
How many coins will the Dandelion-owned nodes get during testnet phase?
This also depends, on both auction results and the duration of testnet, but 5% of total supply is a
conservative maximum
COMPANY
Dandelion, the company is not publicly traded and there are no plans to become publicly traded
at this time.
Dandelion, the coin, is a computational ticket-to-ride on the Dandelion network, and its value
springs from the value of this service to the world. The high speed, low cost, and convenience of
Dandelion transactions means this value easily intermediates trades between counterparties.
Dandelion is built to be used, and every transaction you make adds value to the network.
Dandelion, the network is made of nodes. People interested in running a node can bid for a node
slot during the testnet auction (and we hope you do) but this is not an investment in Dandelion.
Running a node is a competitive business. We will do all we can to set you up for success, but
the results are up to you.
Dandelion, the community, is the best place to create decentralized applications we can imagine.
Investing in building these applications, and the people who build them, is a tremendous
opportunity. The more you use the network, the more value it has, and so the more value
Dandelion coins have. It takes a village to raise a child, and a community to create the
decentralized future.
No. To date we have done everything with peer-reviewed research grants, team investment, and a
small amount of investment from private individuals and the Revtech fintech accellerator. This
approach has allowed us to build value without no pressure to meet artificial deadlines. We are
open to accepting VC if it makes sense for the company, the community, and the network.