Thanks to all who attended Discord Stages on Wednesday where @bux and the team answered some of your questions. For those who missed it here’s a summary.
Quotes and emissions
The initial plan was to distribute approximately 54,000 ANT tokens per day, selecting 100 random nodes every minute and awarding each with 0.38 ANT. Recently, retrieving quotes from the Network has taken significantly longer. As a result, the reward interval was extended from one minute to two minutes, and we compensated by doubling the reward per selected node from ~0.38 ANT to ~0.76 ANT. This reduced the likelihood of any given node receiving a reward, but the issue has now been fixed and we are back to the one minute interval. However, it’s important to note that with up to 40 million nodes, the chance for smaller node operators to receive rewards is still low.
Backwards compatibility
After more rewards testing, we will push an update next week that relates to relay nodes rewards and backwards compatibility. We want to ensure that the network rewards nodes that act as relays for other nodes to connect. This is expected to boost upgrades as you can essentially make more from just upgrading - and we would always rather use a carrot than a stick for such ‘nudges’. Please upgrade!!
Uploads
Not enough nodes have been responding to requests for quotes, leading to uploads failing. The reasons seem to be:
Quoting timeout, which didn’t present as an error to end users but was an underlying cause.
Chunk proof validation, which is where an upload would “succeed” but then couldn’t be retrieved, rendering it useless.
The issue not showing up a serious on our testnets, the hypothesis being that the scale of the network is also a factor (30M+ nodes on main, versus 2K nodes on test). That said, we have been able to reproduce these errors locally, so we have something to go on.
Whales
It has been suggested that we reduce emissions and distribute them over a longer period, to reduce the incentive for large operators. However, we cannot put our thumb on the scale here - someone earns proportionally to what they put into the system. We always prefer carrots to sticks (see above), and once the demand side kicks in with new apps and awareness of Autonomi increases we expect to see the distribution start to balance out.
Partners
Partners like FYEO with their Breach Database of 29 billion credentials, have large amounts of data (30PB) that they ready to upload once the issues are fixed.
Persistence
Once we get uploads working flawlessly we will move to testing downloads. This will require an enormous amount of kicking and punching on the network to determine whether we hit 100/100 or a million out of a million every time, a 1 in a million fail is still a fail in this scenario. Only when we are happy will we be able to state that data is persistent.
REST API?
We’re focused on fundamentals right now, which includes tools for others - specifically APIs, documentation and standards. REST would be a gateway (as it needs a server running it) so it would be something MaidSafe could develop and community deploy, or it could be that a separate commercial entity builds one.
Impossible Futures
We have launched out Impossible Futures programme and a very early version of the website with $100,000 plus 1.5 million ANT available to backers and builders. We are consulting now on the defining the qualifying criteria. It will use an Alpha net - we have not deployed this yet, but it will be deployed ready for use April 22.
Programme rewards
Builders will be awarded with around 50% of the USDC pool (based on number of ‘verified/qualified’ participants). Backers will be awarded with around 50% of USDC pool (based on final builders’ allocation) and 100% of ANT pool, plus ANT incentives for using ANT to vote.
Builders:
- Apply to develop a proof of concept ‘POC’ (on the testnet)
- Compete to produce a working application (on the testnet)
- Qualify for rewards pool via successful ‘verified’ deployment (mainnet)
Backers:
- Post 2 week testnet backers are able to vote for their favourite POCs
- Top 12 POCs that go to next stage can have token locked against them
- At the end of the development window in phase 2 backers can vote (50:50, votes:judges) on the programme winners.
General progress
@rusty.spork is planning more outreach activities, including (hopefully) an AMA in the near future.
@anselme has uncovered a bug with serialisation of XorNames, and made a documented fix around that. This should come in handy for community members using the CLI as an app debugger, as they can now analyse any address thus:
ant analyze 360597a525a9900770c5f6716a0a406a0bea10a2dbe1abe5be9f1b0b0f993851
@bzee worked on Node.js bindings and pushed some updates, but it still needs some more work to set the basics for tests and docs.
@qi_ma raised a PR which seems to improve the quoting performance in the production network significantly, and another to further improve the cost estimation efficiency. He also raised a third PR which during devnet runs, we have encouragingly seen no upload failures so far.
@chriso worked on the latest realease which includes the improved cost estimation and client quoting performance PRs from Qi, on top of many other tweaks and refinements.
In our testing so far, the ‘not enough quotes’ and ‘chunk proof verification’ errors have been eliminated. That doesn’t mean it will hold true on the much bigger production environment, but it should hopefully be a significant improvement, with less time for nodes to supply quotes for storing data. We hope this will translate into quicker and significantly more reliable uploads.
@dirvine has been working through a number of customisations to raw Kademlia that we can experiment with to improve network performance.
Ermine continued debugging the node spawner issue on the CI. The sender and receiver verification is not properly working for the client attempting to connect to nodespawner.
@mick.vandijke has been active on Discord answering queries about gas fee limits in Ant (client) 0.3.8. He explains
A GWEI is 0.000000001 ETH. WEI is just a name for the smallest ETH unit. There is a default limit of 0.2 GWEI per gas of a transaction.
You can see gas like an action point. A transaction on EVM costs a base amount of action points. Doing additional actions inside a transaction (like token transfers) will cost more action points / gas. For the amount of action points you use per transaction, you will have to pay a certain amount of ETH (in WEI/GWEI, which is just a tiny amount of ETH).
If the network is really congested and the price per gas goes above the limit of 0.2 GWEI per gas, the transaction will now fail as of v0.3.8, unless you specify a higher limit using
--max-fee-per-gas <WEI>
As a rough estimate, uploading 4 chunks will cost about 1 million gas.
Mick also raised a couple of PRs [1] [2] and updated the node emission service to retry unexpected failed payments within payout rounds (unexpected fails are insufficient gas fees & gas fee above limit).
Lajos has been looking at the token demand and thinking about adjustments to the smart contracts and awards to incentivise app builders.
@roland fixed a bug that saw testnet nodes dialling mainnet contacts, and worked on changes to automatic NAT detection.
@shu along with @roland and @qi continue to look into node updates and backward compatibility.
And Victor continues to work on the Python client API.
I see you typing 
