Exploring the tangle (part 3)
Very limited information is available about the usage of the tangle for transferring economic value. This is however an important indicator for the performance of iota’s mainnet, one of the most important, I would say. That’s why I started investigating the tangle myself. In this article I will share some the most interesting findings with you.
Data
The analysis below is based on a data set with (almost) all transactions on the tangle (main net) between August 3 and December 13 of this year where value is being transferred. A small part (probably < 1%) of the value transactions did not end up in my data set for technical reasons. (If anyone has any suggestions for a robust way to export transaction data from my node to a readable format like csv, sql or the like, please let me know.)
Before I started working with this data, I removed transactions that were not consistent, that were not confirmed, that effectively did not move value to another address, and which where the cause of address-reuse.
Transactions and bundles
For a good understanding of value transactions on the tangle, it is important to distinguish between transactions and bundles. I will refer to the complete value transfer of iotas from one address to another as a bundle. A bundle consists of separate transactions. A transaction is the value change of a single address. In general I will talk about bundles, because they contain more information about the nature of the value transfer.
Statistics
Over the measured period almost 448000 bundles were attached to the tangle where one or more iotas were moved to another address. That amounts to 3400 bundles per day, or on average one every 25 seconds.
The vast majority of transactions involve only one or a few iotas. These can be qualified as typical test transactions as their value is negligible. In the range between 10 and 100 ki quite a few bundles are transferred as well. All these micro transactions taken together, 90% of the bundles have a value of less than 1 Mi. The bundles with a more serious magnitude seem to be fairly normally distributed around 100 Mi. The exception is a group of bundles with a size between 1 and 9 Ti. This is probably due to the fact that exchanges such as Binance execute their withdrawals by splitting a large number of small outputs from one large input (> 1 Ti) in successive bundles. You can read everything about how Binance handles iota withdrawals in my previous article: https://medium.com/@alexanderhardeman/exploring-the-tangle-part-2-12cfff487fa5. As a result, these withdrawals (small and large) count as large bundles.
Transaction size
Apparently the total amount of tokens transferred is not always an appropriate proxy for the value of the transaction. Especially if a bundle consists of one small output and one large, the question arises which of the two a real transfer is (as in, value traded for something else), and which one the change. I therefore propose to distinguish between two ways to determine the size of a bundle:
- Magnitude: the absolute value of the largest transaction in a bundle. This is the easiest way, and mainly provides information about moving value between addresses, but ignores the fact that often a (large) part of the value remains the property of the sender. It serves as an upper limit of economic activity.
- All values except largest output (AVELO): the sum of all outputs except the largest output. In this case we assume that the transaction with the highest value in the bundle is the change. It probably gives a more realistic picture of economic activity, especially in the case of exchange bundles, but sometimes also misses value (e.g. when the full value of an address is transferred from one party to another). It therefore serves as a lower limit for economic activity.
If we use the second method, the picture shown above paints the value distribution. Intuitively, it seems to fit in well with the distribution of iota across all addresses.
What is particularly striking is that 93.6% of the value bundles no longer have a value when calculated like this, in other words: these bundles only have one output.
This does not mean, of course, that all these transactions are insignificant. It just means that we don’t know whether this is a transfer within the same seed (same owner), or that someone is transferring the full amount from an address to another seed/owner’s address.
By the way: 5.1% of the bundles have two value outputs, 0.7% have three, and so on, up to one bundle with 101 outputs.
Total transferred funds
In the period under investigation, at least 140 Ti was moved (AVELO) but no more than 13.7 Pi (magnitude). This is a wide range, and this illustrates how difficult it is to properly measure economic activity on the tangle. Without additional information about sender and receiver we cannot distinguish whether a transaction is primarily moving between one’s own addresses, or whether it is a transaction that is actually offset by economic value.
Development
Most interesting now would be to see the development of the activity on the tangle. Successful development and implementation of applications in which value transfer plays a role should become visible in the volume of value transactions on the tangle. Unfortunately, there is as yet little evidence of this. In the period August/September there still seemed to be some growth, but in the first week of October both the number of bundles and the value of the transactions decreased considerably (note the logarithmic axes). In November and December, both seem to stabilize again at a lower level, although there is a large day-to-day variation.
The decrease is visible in all segments, i.e. in small and large transactions, and regardless of the way in which the value is measured. Maybe my next article will go a little more in-depth into this.
Another future article will focus on specific types of transactions I found during this analysis.
Do you have any suggestions for further research? Let me know in the comments.
