Dear readers,

as many of you may have noticed, there was a Byteball Airdrop for all Steemit users a few days ago, where you could get Byteball bytes worth up to $160 US (depending on your Steemit reputation).

Since you could even get referral rewards for introducing other Steemit users, the whole thing quickly spread like wildfire. Unfortunately this was a bit too much for the Byteball network and so it stopped working at Friday evening .

_{Screenshot of the Byteball-Explorer}

Background Infos

To understand the causes of byteball network overload, you need to know that byteball has a radically different approach than most other crypto currencies. There is no blockchain:

Byteball pursues a radically different approach: get rid of the blocks! No blocks - no restrictions on the number of transactions. The underlying protocol is called DAG (Directed Acyclic Graph): https://en.wikipedia.org/wiki/Directed_acyclic_graph

Source

This also means that there are no miners, no POW, no POS and no restrictions on the number of transactions. Instead, there are so-called full nodes, which store a complete copy of the Byteball database and witnesses, who check its validity.

Since the DAG chain does not have a secure timestamp (or block number), the peers need a reliable source of transactions that are guaranteed to be generated in a defined order. This prevents transactions originating from a fraudulent fork (a so-called shadow DAG) from being confirmed. In addition, you can recognize double-spends (double payment with the same coins).

Source

The Problem

This has apparently always worked smoothly, but as I learned in the Byteball Slack, there was probably a problem with the database because they decided to allow transactions that have not yet been confirmed by the witnesses. This has made the DAG very complex.

_{Anton "Tony" Churyumoff (Chief Developer)}

A user named Peter Miklos (who runs a node and also the Byteball Market) mentions, that "it sounds like the spending unconfirmed funds feature just accelerated the scalability issues that have been already there before".

He continues writing:

Considering that the chief developer not only seems to be the only one who can read the code, but also runs all 12 witnesses, all this (despite many interesting approaches) does not make the most trustworthy impression at the moment.

There are 12 witnesses in Byteball and the default setting in the wallet currently refers to 12 witnesses all operated by Byteball chief developer Tony (Anton Churyumoff). As airdrops are still being carried out and a large number of coins are still in Tony’s hand, this is acceptable. In the future, however, these witnesses will be operated by a more varied group of users. And it clearly makes sense that those users disclose their identities because of the position of trust. But the time is not ripe for it yet.

Source

TL;DR

Like most other Steemians, I was excited at first that Byteball decided to use the Steemit platform for an Airdrop, integrating both the Steem reputation and Steem usernames into their system.

Of course, it is disappointing that the Byteball net collapsed one day after the announcement of the Airdrop. Although the chief developer will certainly do everything to solve the problems and stabilize the network again, I think it is a great pity that the problems were not communicated openly on Steemit and that they didn't eventually asked for help.

Meanwhile, all transactions are slowly being processed and I hope that the Byteball Wallet can be used again in a few hours. Of course, I will also continue to refer new users by sending them Bytes, so that they can authenticate their Steemit username and get their free Bytes.

If such a centralized system, as the Byteball DAG currently seems to be, deserves your trust in the long run, however, is totally up to you.

Links

Here are some interesting links for you:

Very interesting interview with the Byteball community manager "CryptKeeper".

Part1
Part2

The Byteball Explorer, where you can check the actual state of the Byteball database.

The Byteball Slack, where you can join the discussion or offer your help.

Thank you for reading!

@shortcut

---

Update (07/16): The Byteball chief developer Tony Churyumoff just explained, what happened in the Byteball Slack:

• on Friday (the 13th) 10 out of 12 witnesses crashed due to a bug. The 2 surviving were recently updated and didn't have the bug
• among the 2 surviving witnesses, was the steem attestation bot which continued to post transactions very fast, and since it didn't have to wait for confirmations, nothing could stop it and the unconfirmed tail of the DAG has grown very large
• the 10 witnesses updated and restarted but they started adding their units to the old part of the DAG they last witnessed, which caused massive rebuilds of the main chain each time a new unit was added. This made validation of new units very slow.
• I had to urgently find and apply a few optimizations that matter only when the DAG takes such unusual form. They made validation faster but still way slower than usual
• when the optimizations were ready, another issue popped up: calculation of stability point gave unexpected result. I had to investigate the new issue, which took another day, and found a mistake in the algo. The fix breaks compatibility with the old (flawed) algo, and makes the update mandatory for all full nodes.