Repository

https://github.com/FutureShockco/drugwars.io

Introduction

Drugwars is one of the most widely played blockchain games not just in the Steem blockchain but also in all blockchains. It was a huge hit in the beginning, but now revenue is getting down.

There are some major changes so far.

• Drugwars wanted to change the reward scheme with FUTURE tokens, but that plan (without enough grace period) was harshly criticized and then postponed.
• Fighting against macro/bots, custom_json transactions are now encrypted.
  (Due to this structural change, detailed analysis will not be available anymore, and hope they will be some way to verify the transactions, as I suggested in Drugwars as a blockchain-based game - verifiability and irreversibility.)
• Activity rules (inactive users for two weeks will not be rewarded).

Despite the roller-coaster ride, Drugwars is still No.1 in the State of the Dapps rankings as of today. Thus its data is important not only for Drugwars but also for other dapps.

Scope

• Data: 2019-02-11 - 2019-04-07 (steemsql)
• All transfer & custom_json transactions are analyzed

Note: there can be invalid or duplicated custom_json transactions. This is mainly due to bots or bugs. Since there is no way to check which transaction is actually accepted, I count all as valid transactions. Of course, there are some obviously unacceptable transactions, but there are not many, and you can still see some invalid transactions after transaction format (encoding) changes in the plots.

Note: there were some best-test period before 2019-02-11, so some cumulative numbers may be different from the exact data.

Results

Revenue, Reward, Profit

While I didn't play at that time (I started quite late and I've been just playing for the test/analysis purpose from time to time), I knew this game when it began. Only after a week, it got 100K STEEM profit! While 100K STEEM isn't a big money, considering the limited number of STEEM users, it was a hugely successful debut.

However, unfortunately, the cumulative profit (red) is going down after hitting about 130K STEEM in less than 2 weeks. It's better to see the following graph without the cumulative profit to see the daily revenue stream.

• Note that two peaks in reward is due to events in the game.
• There are also some refund. While I subtracted from the profit, it's not that big, so I don't include it in the plot.

I was also surprised to see that only after 13 days, the daily profit became negative. And except for two almost negligible days, the daily profit has stayed negative. And the reward is also decreasing (as you know if you're playing the game).

Number of Users

We all know that there are many alt accounts. While figuring out how many alts exist is interesting, but it's also very difficult. After all, we are in the blockchain world with pseudo-anonymity. Thus, you really should read the following numbers as the number of accounts not users.

• All: cumulative users
• DAU: daily active users (who makes at least one transaction)
• New: daily new users
• All_paid: cumulative paid users
• DPU: daily paid users
• New_paid: daily new paid users

DAU is more than 6K which is quite nice, considering that Steemit daily active posting users is slightly less than 10K.

One very interesting fact is that DAU is very similar to cumulative paid users (All_paid, red). Once they paid, they play hard to make a profit :)

From the above plot, the following plot shows New/DPU/New_paid only.

• After a couple of major adoptions, new sign-ups are decreasing.
• Daily paid users and new daily paid users are also decreasing.

While 6K DAU is quite decent, we all know that there are many alts. Thus another good measure of real active users is number of users who received the reward, which will be shown in the reward analysis section.

Revenue Analysis

Except for the very recently introduced FUTURE token ("shop") purchases, the revenue consists of upgrade and unit recruiting.

One very surprising thing to me was non-negligible people are spending money to get units. In the early days, many of them might be test transactions. However even after those period, some users are still paying for the units. Based on my calculation, it is hard to make a profit from buying units in most cases, so I interviewed :) one user who spent a decent amount of money on units, and his said it was an impulse purchase. He was in the middle of a big fight which he didn't want to lose, so he just bought them and then regret :)

I actually found one well-known user to most Steemian, who spent really a lot of money on units, and I'm really curious about the intention and profit. Anyway, that's why in terms of the number of purchases, upgrade is much dominant. (There are only a few users who are spending quite a lot on units in one transactions.)

• People spend more on Friday and Sunday. Friday might be due to TGIF and Sunday might be due to defense preparation for the upcoming week :)

Reward Analysis

• Note that two peaks in Heist reward is due to the events in the game.
• Except for the initial period, the total amount of Daily rewards and Heist rewards are equally divided, as you can see in the game.
• Due to the decrease of the revenue, the reward pool is also decreasing continuously.

Number of Rewards

As I mentioned in the number of users section, the number of rewards can be seen as a measure of active users.

Note that Drugwars pays even when a user doesn't play at all up to two weeks. But since the encoding is introduced (while some may break the code), at least the number of Heist reward transactions (which is provided when users sell drug, which need manual interventions.) can be considered the lower bound of the real active users.

• At least 4K accounts are quite active. Again, due to the encryption introduced on 03/25, the Heist reward should be done manually.
• Step-wise decrease of Daily rewards is due to the decrease of daily reward for the same level.
• The number of referral rewards can be seen as the lowest bound of the real users. While it is still possible that some users signed up without referral, most smart users who actually play the game would open another account to get the referral bonus. Of course, it is also possible that one user created multiple referred accounts.

• Due to the stable number of rewarded users, the average of reward is also decreasing.

Transaction Analysis

Except for the transfer transactions for purchases, Drugwars is using custom_json transactions for other transactions: sign-up, non-paid upgrade, unit, fight, heist. Therefore, together with transfers, custom_json is a good source to see user activities. Due to the introduction of the encoding, it can also be used to find out accounts who are still using macros which are not allowed anymore.

• drugwars: Until 03/25, this was only for fight, but since 03/25 it's for all transactions. Basically users cannot tell which transaction is which, which I believe not a good thing, as I suggested a way to resolve this problem in Drugwars as a blockchain-based game - verifiability and irreversibility
• dw: This was temporarily used only for a few days, when the encoded transaction was initially introduced.
• dw-char: sign-up
• dw-heist: sell drugs (which gets Heist rewards)
• dw-unit: unit recruit
• dw-upgrade: upgrade buildings
• token: purchase FUTURE tokens (very recently introduced, can be ignored now)
• unit: paid unit recruit
• upgrade: paid upgrade transactions
• total: sum of all transactions.

Again, "drugwars" was initially for fight only, so you can only see from when the fight system was introduced. However, since 03/25, all transactions have "drugwars" id, which is not good for analysis.

Initially, the number of fights was smaller than the number of heists, after a certain point, it is getting larger. My conjecture on this is, Heist is enough once or two per day, but fights can be done many times per day. Unfortunately, due to the encoding, we cannot see the recent trend anymore.

To see the ratio of paid transactions easily, I show the following plot.

• In terms of number of transactions, the ratio of paid transactions is small and getting smaller.

Conclusion

• Drugwars may need some changes to survive. One positive thing is that the team seems working hard. There have been many changes recently.
• As I mentioned in Drugwars as a blockchain-based game - verifiability and irreversibility, even if users cannot get a profit, they cannot blame the game. You should admit that it's not a principal-protected financial product, but a game. You bought an item in the game. However, the Drugwars team also should make an effort not to misguide people, and also provide an enough notice in advance for any big change.
• As I also suggested in Drugwars as a blockchain-based game - verifiability and irreversibility, Drugwars should provide a way for transaction verifiability for being a real blockchain-based game. Hope @hightouch and @fabien will resolve this. Thanks.

Tools and Scripts

SQL query and python scripts are available on my Github:
https://github.com/economicstudio/steemanalysis

Relevant Links and Resources

• Drugwars as a blockchain-based game - verifiability and irreversibility by blockchainstudio
• How much automation is on Drugwars with scripts and bots? by crokkon

My other analysis articles

• Witness of the people, by the people, for the people
• 100% SP vs 50:50 Which one is better? Does the haircut matter?
• Effect of haircut, early voting, beneficiary on dust payout
• Decline of busy.org posts - Need power up for newbies!
• Utopian bot VP analysis - 2 more rounds are available per year :)