Valve’s “Secret Weapon”

Valve’s Portal was nearly scrapped during its development due to playtesters’ feedback that they couldn’t distinguish between the tutorial and the actual game. This led Valve to create GLaDOS, a witty AI antagonist that became one of the most iconic video game villains of all time. Valve’s approach to game development is heavily focused on playtesting, which involves watching players interact with the game and using the feedback to improve the design. Playtesting helped Valve touch up almost every aspect of Portal, from the learning curve to the iconic visual style of sterile white walls and floors. The obsession with playtesting can be traced back to Valve’s disastrous development of their first game, Half-Life, which was nearly scrapped two months before its release due to serious technical problems and poor level design. The lessons learned from Half-Life led to Valve’s two game development philosophies, including using small multi-disciplined teams and frequent playtesting from the very earliest point of development. Playtesting led to numerous changes in Half-Life, such as stuffing crates with goodies and implementing the x-ray outlines of fellow survivors in Left 4 Dead. In conclusion, Valve’s obsession with playtesting has led to the creation of some of the most iconic and successful video games in history.

Related Articles


Your email address will not be published. Required fields are marked *

  1. This was my literal experience with portal: "What a cool tutorial, and a great segue from tutorial to game by making me reach the surface. I can't wait to apply this portal stuff combined with weapons here in the actual – …" credits roll.

  2. What I want to know is how many people playtested each week; we know the regularity and how early they start, but not how many people they test each week, and I'd find that information to be valuable too

  3. Love this video! As someone who's done voluntary playtesting for a different large game company, seeing Valve's process is interesting. When I've playtested, the company would give screening surveys beforehand, then select the people they wanted who had attributes they desired. I've tested with a handful of groups, from female gamers to a room full of people who had no idea how to play a certain game genre. Online playtests and post-release playtests are certainly a thing, but nothing beats being in a room with a playtesting team.

    For anyone wanting to get into playtesting that have the ability to do so, here are a few tips:
    – Listen to the devs (sounds obvious; if they say to play their game for only half an hour, only play it for half an hour)
    – Play the game as normal, but keep notes in the back of your head (or even write them on a piece of paper, if given one). Especially note anything that confuses you, looks weird (specifically in relation to stuff like the UI; if the textures look odd, no need to mention it), or just plain doesn't work. The playtesting process is there to help the devs work out problems, and they can't really do that if the person testing goes 'yeah it was good' and then leaves.
    – Ask questions! Sometimes game developers don't know that x confuses players (so they don't do anything about it), and the player doesn't know if x is intentional or not (so they ignore it and move on). It's best to communicate more instead of make assumptions about the product.

  4. Take a look at boardgame playtesting articles. Boardgame designers do this very well because they have to – there are no patches. This is one of the most robust elements of the boardgaming community with play test sections at every meaningful board game expo/convention.

  5. Very good video. Lots of designers need to do this. You don't need Valve funds to do it. Boardgames do it all the time on a shoestring budget.

  6. I worked in a company that put a lot of effort into testing ideas. Here is how testing looked from the perspective of us devs:
    – It looked like the leaders didn't trust their ideas, didn't have coherent vision and direction
    – It felt unfair – we devs are expected to work lightning fast and implement the craziest thing but managers are allowed to spend months just to make a simple decision
    – you have no idea how to structure your code, because the vision today will not match the vision tomorrow. So your code is just a pile of spaghetti

    My point is that testing is amazing, sure, but can your company do it right, before you go insane ?

  7. The downsides of valves style is how quickly they reject design ideas and how slow they produce workable peoducts.

    Resulting in difficulty find direction that satisfies the group developers.

  8. What crazy idea to release this video before the GMTK Game Jam! It would surely have its fun influence on the Jam! I will be so excited to see many quirky games!

  9. "Each game in the series features a different piece of technologies developed at Aperture Science" Sounds like what Nintendo did with Zelda BotW and TotK… and it worked out pretty well for them!

  10. I mean, this all sounds nice…but it still doesn't explain why valve is allergic to making a third in a series…. by now there's no way that HL3 can live up to the hype, cause we've been waiting for it for over a decade :'( but I still want it 0.o
    and it's not just half life, half life has 2 main games and some episodes that are basically expansions then there's portal which has 2 main games, and left 4 dead has 2 main games because what would be the third got rebranded into back 4 blood.
    these are 3 of their most well known, iconic and loved franchises. yet they feel underused. it's obviously better than going the fifa/CoD route and going full greed mode by pumping out barely improving versions year after year. but still, shouldn't be that hard to give us a new full length half life game.

  11. What I'm curious about here is… where do they get these playtesters and how do they keep them unbiased from the development process?

    Because conceptually it's great… but practically it's a lot to get that many players to try out your stuff.

  12. Amazingly well prepared episode. Thank you. It just made me think of the `Lean Startup` paradigm developed by Eric Ries and presented in the book of that title. It talks about `build-measure-learn` feedback loop as the primary tool for development of any product in any business. It was conceived to protect entrepreneurs from building a product that nobody wants by testing assumptions early and adjusting to feedback. The only thing that doesn't change in this process is the vision, which, as mentioned in the video, keeps us from providing a 'bland, designed by committee, sludge'.

  13. Valve has the worst play testers in the world. they would get better results and amore intelligent response letting a rock playtest than their play testers.

  14. 11:20. LOL. I'm taking notes as I go. I paused the video 10 seconds earlier and wrote my notes.

    "Yes, even if they're stuck for 20 minutes. It is frustrating seeing them miss the "obvious" answer, but seeing how far they go and what they do is more important".

    Then resumed and saw you practically said the same thing, down to the exact minutes and quotation marks on obvious. Kind scary/cool.

  15. The idea with a camera sounds like a possible approach for younger audience. Kinda sorry that they've scrapped it after a year of Dev ^^'

  16. I didn't know that concept used in Superliminal and Viewfinder was originally supposed to be a Portal sequel.

  17. glados is so amazing the main reason I played portal as someone who does not play any full puzzle games

  18. The cabal seems a lot like cross functional squads in agile methodologies. The iterative testing and feedback cycles feel very agile as well

  19. This honestly explains a lot about Valve's development times. No wonder it takes them so long to make a game, they literally spend half their development time testing it!

    Not judging, to be clear- it's pretty obvious that this approach does work for them.

  20. This games designer for Portal Kim Swift after portal then went on to design another great puzzle game that I feel people don't talk enough of is Quantum Conundrum. Two games that in my mind revolutionized and created the new genre that is First Person Puzzle Platformer.

  21. Wait, are you trying to tell me that genuine care for what you create and the people you're creating for, results in a game that is not at all lacking in soul and creativity? Who could have predicted this?

  22. As an admin to one of the biggest csgo mapmaking servers, I can assure you that playtesting is exactly what takes a map from a cool idea to a working map.

    The maps that get into the game are sometimes playtested for over 100 hours with other mapmakers.