“What’s up with the Tangle?”

This is a common question on our Discord and here on Reddit. Questions about the state of the Tangle usually stem from the way it is represented in a visualiser or due to a high/low confirmation rate. I’m going to quickly walk through some of the questions asked and finish with a brief look at the current questions.

When ever these questions are asked in the community, we are asking them ourselves. We are all as intrigued as everyone, if not we’d be doing something else. These events present learning opportunities for those working on IRI and other aspects of the network. As developers test software on the Tangle we are able to glean further information on the mechanics of the protocol and how it reacts to stimuli that we couldn’t dream of in a simulation.

Let’s first take a look at what I think are two main reasons people have been asking about the Tangle recently:

TPS is King

A lot of the community is interested in pushing the TPS higher and higher on the mainnet. This is a straightforward metric for comparison with other DLTs and allows us to rank them neatly in our head.

Our desire to see the Tangle “outperform” other networks leads us to spam the mainnet in order to push the TPS as high as it can go. Without fail this results in ‘Wow, the tangle is flying today,’ and then ‘Why did the TPS suddenly drop’ after someone turns off a spammer.

This is a noble goal but it shouldn’t be the yardstick we use to gauge the “state of the Tangle” at any given moment. For me, spamming is inconsequential compared with the thing that will push the Tangle forward: adoption. Adoption will naturally spam the Tangle and maintain a high TPS without artificial drops. Even though adoption cannot be measured with one simple metric, it is nonetheless a much more important metric that one must gauge when considering whether or not the tangle “works.”

Misunderstood Metrics

Generally, questions stem from a focus on TPS rather than CTPS.

  • TPS (Transactions Per Second) — This relates to the raw number of transaction published to the network per second.
  • CTPS (Confirmed Transaction Per Second) — Relates to the number of transactions that move from pending to confirmed per second (averaged).

A common question is: “Why is the CTPS so low compared to the TPS?”. In a perfect world, the CTPS should almost match TPS, considering an optimized Tip Selection algorithm being used by the vast majority of network nodes.

When someone starts sending bad transactions which outweigh the honest transactions by 5x the confirmation rate will drop to 20%. In reality the confirmation rate didn’t drop, the amount of invalid or poorly referenced TXs just increased. The honest throughput of the network remained the same.

What is `X`? Maybe it’s `Y`

Often when people ask a question they will speculate on what the supposed issue is. “I think it’s Satoshi reaching down” or “A rival project doing X”. Unsubstantiated speculation doesn’t contribute any new information to the situation.

I applaud the small group of developers who do jump in and try to understand what the situation is and provide considered information back to the community and core devs for improvement of the protocol.


Not to beat around the bush for too long, this post was brought about by the recent occurrence of a ‘Side Tangle’ appearing in visualisers.

What is a Side Tangle?

Side tangles, as they have come to be known, are actually a well-known phenomenon since the Tangle white paper, which refers to them as parasite chains. Parasite chains, like this one, are the result of a spammer using a modified software to select tips that only reference themselves. This results in a tangle of transactions being built up that is distinct from the main tangle.

The effect of this is functionally equivalent to choosing a bad strategy for tip selection: transactions in the parasite chain are less likely to be confirmed, and that is exactly what has happened in this most recent case. The Tangle is performing as theorized in the white paper!

When the recent IRI update was released, the Tangle was looking really “healthy,” with a 98% Confirmation rate and an average sub-2 minute confirmation time. CTPS was over 20!

However, all good things come to an end and someone started spamming transactions with a low likelihood of confirmation (‘Side Tangle’) and the metrics took a nosedive. To speculate, reasons could include: attempts at price manipulation, testing network response to different things, or just to get attention.

Given this drop in our “go to” stats, people have naturally been asking what this all means. The questions have even led to​​ memes:

Do side tangles actually rock…

…and should I expect the unexpected?. Maybe, we’ll find out sooner or later.

IOTA is a permissionless system, which means that anyone can create a node and send transactions, whether they are honest or malicious. The protocol has been designed to be resilient to funny things happening.

This isn’t the first time people have sent transactions with confusing tags and weird tip selection. It won’t be the last time either.


We as a community should observe and critically evaluate, just as we are are constantly doing within the Foundation. Only through constant learning and incremental improvement can push the IOTA project forward.

Like what you read? Give Lewis Freiberg a round of applause.

From a quick cheer to a standing ovation, clap to show how much you enjoyed this story.