This week a developer published a teardown of how Block Blast, one of the biggest casual puzzle games on the stores, reportedly decides which ads to show, when to offer a revive and how hard each round should be. The findings were picked up by trade sites and newsletters within days, and they have split opinion between “clever monetisation” and “a rigged game”.

Both readings go further than the evidence allows. This post sets out what is reportedly in the code, what has not been shown, and which parts are useful for anyone designing monetisation or buying installs for an ad-funded game.

What was reportedly found

The analysis comes from an independent developer, Ayo Omolewa, and the details below are relayed through a trade-press summary and a news report on the difficulty findings. We have not seen the code ourselves, so treat everything here as reported, not confirmed.

  • Most of the game logic is delivered by the server. The installed app is largely a shell. On launch it reportedly receives a configuration naming hundreds of small modules, each controlling one part of the game, and part of that code is downloaded after install. Two players can therefore get different rules, and rules can change without an app update.
  • A small model on the phone predicts what each player is worth in ads. It reportedly takes about 27 inputs, including recent ad prices, country, device model, memory, install source and how the first game went. In one test with a made-up player, changing the country alone cut the predicted value by roughly two thirds.
  • The ad format follows the prediction. If rewarded video is predicted to pay at least double, the player is reportedly offered extra revives, each backed by a rewarded ad. If interstitials pay more, the extra revives are not offered, and players predicted to be worth very little may not see interstitials at all.
  • Hands that cannot be placed. Several rules reportedly make a round unwinnable, mostly just after a player beats their personal best, which is when a revive is worth the most. One rule is described as giving each new set of pieces a one in five chance of being impossible to place once a player is ten percent past their record.
  • Safeguards exist too. After two bad games in a row the player reportedly gets easier pieces, and if a model predicts they will play less tomorrow, the next revive may be free with no ad.
  • Everything is measured. The code reportedly tests different timings for when an unplaceable hand appears, and checks whether players start another game within fifteen minutes and how often they accept revives.

What it does not show

The teardown is one person’s reading of encrypted, server-delivered rules, and the newsletter summary adds its own interpretations on top. Several points are open:

  • The reports do not say how often the rules fire, or whether every player is affected.
  • The model examples come from a single made-up player, so they show how the model behaves, not what typical players experience.
  • No response from the publisher has been reported.
  • Revenue figures going around, including a daily estimate near the seven-figure mark, are extrapolations from third-party audience data. They are not disclosed financials.
  • Labels such as “rigged” are opinions. Deliberately adjusting difficulty to hold attention is a long-standing design practice. Where that stops being balancing and starts being manipulation is a judgement call, and it is exactly the line regulators and the stores care about.

What studios can take from it

Setting the ethics debate aside for a moment, the reported design points to several things that apply to any ad-funded game.

Players are not worth the same. Country, device and install source change what a player’s ads earn, often by a large factor. If your reporting averages everyone together, you are hiding the differences that decide whether a campaign pays back. We track value by source and country for this reason, and it is the first thing we look at in a growth audit.

Format choice is a lever. Rewarded and interstitial placements compete for the same attention. Pushing one harder changes the session, the retention curve and the revenue mix. Test the mix, and judge it on retention and ARPDAU together, not on ad impressions alone.

Server-side control is powerful. Being able to change rules without an update lets a studio run many small experiments quickly. It also means you need a clear record of what is running for which players, or you cannot explain your own results.

Measure what the friction costs. The most interesting detail in the reports is not the difficulty spikes, it is the safeguards: easier pieces after bad games, free revives for players about to churn. Any design that adds friction needs a matching rule that protects the long term, and a test that proves the long term is not being traded away.

Assume your logic can be read. The reported rules were encrypted and kept out of the store build, and someone still extracted and published them. If your monetisation design only works while nobody understands it, it is fragile.

The trust and policy side

Studios should expect more scrutiny after stories like this. Players talk, and a game that feels unfair loses reviews, which lowers conversion on every paid install. The stores and regulators also have rules on deceptive design and on how ads are presented, and those rules keep tightening. A monetisation system that you would be uncomfortable explaining to your own players is a risk, whatever it earns this quarter.

A practical test is to ask whether a player who knew exactly how the revive offer works would still feel the game was fair. If the answer is no, the design needs another look.

The short version

The reported design is not one trick. It is a large set of small rules, each tested, with a prediction of player value deciding which ones apply. The lessons that travel are to measure value per player and per source, test ad format mix against retention, keep a record of what is running, and build in guardrails. The parts that do not travel are the ones that depend on making players lose on purpose.

For the design side of ad-funded games, see our guide to monetisation design, and for how value per player feeds acquisition decisions, read UA best practices.

Sources: the reports linked above, dated 20 to 23 October 2026. Details are as reported by those outlets and have not been independently verified.