For episode 104, the show’s two year anniversary, Al is joined by Nuno, Adam and Drexy to give their verdict on this October’s Steam NextFest. With over 7,000 demos, widespread AI generated filler and weak tagging, the team debate whether democratisation has gone too far, how Valve could curate better, and which genuinely good demos they still found.
https://criticalmovespodcast.com/listen
This anniversary episode finds the hosts in a grumpy mood about Steam NextFest. They describe a festival swamped by thousands of demos, many of them visibly AI generated, with poor tagging and recommendations that surface games with no connection to strategy. The discussion separates AI assisted development from low effort AI slop, revisits last episode’s conclusion that democratisation is a good thing, and looks at the practical limits of Valve’s upfront fee, verification checks and mandatory AI disclosure tag. The hosts trade ideas for curation, from a revived Greenlight to community AI tagging, before finishing with the demos they would still recommend and a thank you to everyone who has been part of the show.
Critical Moves Podcast: Episode 104 Show Notes
Episode Title: Steam NextFest or Steam Slopfest? Two Years of Critical Moves
Hosts: Al, Nuno, Adam, Drexy
Episode Length: ~60 minutes
Episode Summary
Marking two years of the podcast, Al, Nuno, Adam and Drexy sit down to talk about the current Steam NextFest and agree that it has been a disappointment. The conversation covers the sheer scale of the event, the flood of AI generated games, and the difficulty of finding quality titles through Steam’s tags and recommendations. The hosts revisit the previous episode’s argument that democratising game development is a good thing, ask what Valve can realistically do about curation and disclosure, and close with the NextFest demos they are still looking forward to.
Two Years and Where It Began
Al marks the 104th episode and two years of the podcast, welcoming back co-founder Nuno. He recalls that the show began with an unsolicited email from an English stranger to a Portuguese writer, saying he liked his work and asking if he wanted to collaborate. Drexy, with some humour, takes credit for all 104 episodes.
From Brazilian Batman to Slopfest
The hosts recall covering the first NextFest on the podcast, including a game they nicknamed Brazilian Batman, whose heavy use of AI was their major criticism at the time. They note that NextFest has always been a source of fun for the show, but that this year’s event felt different. Al calls it Steam Slopfest rather than Steam NextFest, and the others broadly agree.
Too Many Demos to Browse
Nuno points out that the festival began at around 7,000 titles and had grown to over 7,200, making it impossible to browse. Even narrowing by a tag still leaves around a hundred games to scroll past. Adam, normally a big fan of fests and demos, says he gave up after searching for tower defence and turn-based games and finding too many near identical titles. Drexy notes that a personalised feature was added a couple of days in and made things somewhat better, though the volume of AI generated entries was still startling. Al says that apart from a handful of legitimate looking titles such as Tropico 7 and Stronghold 4, he found little worth his time and warns that if this continues he may not cover the next NextFest in February.
AI Assisted Versus AI Slop
Al stresses a distinction between AI assisted development and slop, referring back to the earlier interviews with Dan and the episode with Ben. Many of the titles he saw shared the same art style, UI and capsule art, which suggested they were thrown together in weeks to chase NextFest sales. Nuno agrees that there are valid uses of AI in development, but argues that slapping something together with AI and hoping for the best is another matter.
Asset Flips, Greenlight and Democratisation
Drexy likens the current flood to the Unity asset flips that followed the end of Steam Greenlight, where cheap games were assembled from store assets and posted in volume, only now the starting point is an AI prompt. Nuno and Adam say Greenlight should return in a different form, though Adam recalls that votes could be bought from dubious sites. Asked whether last episode’s positive view of democratisation was wrong, Drexy says no, and expects a period during which this kind of content makes up the majority of Steam before something changes.
What Can Valve Actually Do?
Adam proposes a one time publishing payment of a few hundred dollars, plus some form of game verification, to deter low effort submissions. Al responds that a fee of $100 is already required to create a store page and that Valve does check games, mentioning that a project he works on had to delay a launch by a week while waiting for Steam to approve it. Adam counters that reports of malware on Steam suggest checks are patchy. Drexy suggests outsourcing curation to a group of game websites, or taking a more invitation led approach like the smaller themed fests, while Adam argues that a company as profitable as Valve should invest more in tagging and filtering.
Tags, Misleading Screenshots and Odd Recommendations
The hosts criticise the tagging system, where developers add around twenty tags to reach as many people as possible, and note that the recommendation engine pushed unrelated titles, including a dating sim, to the top of the strategy listings. Adam says that some listings use screenshots that do not represent the gameplay at all and wonders whether that is even legal under European consumer rules. Al adds that he could not view some of the games the team were laughing about because of age verification requirements under the UK Online Safety Act.
Legitimate Developers Lost in the Noise
Al says his sympathy lies with the developers who spent their own money and sleepless nights on a game that might be excellent but will never be seen. Drexy relays a friend’s experience of waiting a month for Steam to approve a legitimate game while far lower quality titles appear to get through without trouble, and wonders whether there is a back door.
The Shogun 2 Remaster Controversy
Al raises the re-released Total War: Shogun 2, which was pulled after backlash over its suspected use of AI upscaling. Nuno argues that this is a different problem, a poorly executed upscale rather than a deliberate flood of low quality games, and notes that demos must be approved before they can take part in NextFest, though the criteria remain unclear.
Three Kinds of AI Game
Adam sorts the AI heavy games he tried into three groups. The first are AI assisted projects where a studio has clearly put real heart into the design, the second are AI generated games made by people who really want to make something but lack the experience, and the third, which he says is the largest, are cash grabs built in a day or two with no discernible idea, such as a water themed simulator. Al points out that judgements about heart are subjective and asks how a small team like Valve could police them.
Steam’s AI Disclosure Tag Does Not Seem to Work
The hosts note that Steam requires developers to disclose AI use, but several games that obviously use it do not say so. Adam singles out a Tetris, roguelike and tower defence hybrid called Guns, Princesses and Braves as the best game he played during the fest, praising its carefully balanced mechanics while noting that its art is largely AI generated and the store page makes no mention of it. Al and Drexy say they can spot AI built games from the default UI frameworks used by coding tools. Adam argues the disclosure tag is almost meaningless because virtually every game now uses AI somewhere in development, and suggests a community led tag or an AI meter. Others worry that would be easy to abuse. The group suspects the tag exists mainly to satisfy EU rules, that Valve has little incentive to enforce it, and that without penalties there is little reason for developers to comply.
Placeholder AI and Games Doing It Honestly
Drexy highlights Steel Company and Steel Wolverines, both inspired by Panzer General, as examples of developers who are open about AI use. One states that its AI art is a placeholder, promises a lot of manual editing, and says early sales will fund hand drawn sprites for every unit. Drexy also describes a friend who plans a Kickstarter to hire a proper artist if there is enough interest. Adam suggests an early access style Q&A in which developers answer a few questions about how they used AI, though Al argues that is as hard to enforce as the early access roadmaps themselves.
Al’s Position
Al restates that he has no problem with AI being used to make games as long as developers are honest, but says this NextFest has coloured his view, because the sheer quantity of default UI and Claude coded filler means good games may be going unseen. He teases an upcoming episode on the early accessification of strategy games.
Games Worth Playing Anyway
Drexy recommends Steel Company, Steel Wolverines, Arch Rebel (MicroProse, inspired by XCOM and Laser Squad, with 16 bit era graphics), Veterans Napoleonic Wars, Captive 3000 (inspired by the old dungeon crawler Captive) and Series Universalis, a grand strategy game with a Romance of the Three Kingdoms feel. Adam adds that he has wishlisted Arch Rebel and recommends the Stronghold 4 demo, which he describes as close to a 3D remake of the first game with some bugs, plus Guns, Princesses and Braves, Dawn Defense (an RTS and tower defence hybrid) and Trailblazers. Al recommends Red Recon 1944, a hardcore Commandos style tactics game played from the Soviet side. Nuno backs Drexy’s picks after playing Steel Company and Steel Wolverines over the weekend, and adds False Echo (a submarine game), Stronghold 4, Theatre of War Rearmed, and Veterans Napoleonic Wars, which he says has improved since an underwhelming earlier build.
Thanks and Sign Off
Al thanks Nuno, Adam and Drexy, as well as past contributors Tim, Jack, Joe and Shane, and the audience. The episode ends with a joke about thanking the fanboys, following a running gag from earlier in the discussion.
Contact & Links
About | Contact | Meet the Team | Get Involved | Forum | Episodes
Discord | Reddit | Twitter / X | Facebook
Instagram | Twitch | Steam Group | Steam Curator
YouTube | Spotify | Apple | Amazon
Email: [email protected]
Episode Verdict
This is a cheerfully grumpy anniversary episode that works because the hosts push back on each other rather than simply agreeing that NextFest was bad. Al’s reminders that Valve already charges a fee and verifies builds, and that any disclosure rule depends on honesty, stop Adam’s and Drexy’s proposals from going unchallenged, and the unresolved question of how anyone could prove a game used AI gives the middle of the episode real substance. The three category breakdown of AI games is a useful way of thinking about a problem that is often reduced to a single label, even if the line between heart and cash grab stays subjective. The closing list of demos is a welcome reminder that good games still exist underneath the noise, though the hosts admit that most of them were titles they already knew about, which rather proves their point about discovery.
Discover more from Critical Moves
Subscribe to get the latest posts sent to your email.