I read the entire symposium and wrote a big old wall of text while I was banned and forgot to post it. I would normally spread stuff like this out over several posts but this thread is long dead. So here it is:
The core problem with game reviews:
The core problem with game reviews is that they must serve many purposes for their audience but they are not optimal for any of them. They are used as a buyer's guide, a critique, a recording of how a genre is advancing and how much the bar has been raised, a record of the impact a game had on the community when it was first released, a way to compare similar games, etc. But in the end they're just someone's opinion and they should only be judged and presented as such. That's why the "What you've been playing" segments on podcasts are so refreshing. They are honest and completely unpretentious. I think someone should explore better ways to take care of those other tasks (like a giant game wiki of some sort).
Why review scores do matter
I disagree with the idea that scores are only fodder for dirty, unwashed masses and forum goers, and that reviewers should aspire to rise above them. They can be a good summary of a reviewer's experience with a game. Sometimes that summary is all you want, and no site has come up with anything that is as pithy as a review score yet. I'm not going to read a wall of text if I only want to know if Cunt Kicker 2 lived up to the original before I buy it. Also, words alone are not always sufficient, because there are many aspects of the gaming experience that cannot be fully articulated. The review might even be better off without throwaway qualitative statements (i.e. this game is awesome/horrible). I just think the scores should be given context.
As it was mentioned before, the score can act as a centerpoint for discussion, and it can quickly give you a rough idea of whether a game is worth buying. But again, context is important. Budget games rarely get good scores, and if the premise sounds good enough you can ignore the score altogether. EDF 2017 is a good example of this. GTAIV is on the opposite end of this spectrum. That is why review score aggregating sites have the potential to paint a false picture of a game. The scores are being aggregated to give a result that is devoid of any context. My problem with meta-critic is that the context is just as important as the content (if not more so). A score by itself can lose a lot of its value. Background information on the review process would also help.
The Vanilla ice cream principle: Why popular things tend to be bland
Another thing I've noticed is that the blander something is the more potential it has to be popular. The most popular ice cream is vanilla because its flavor is rather inoffensive, so it will usually be in everyone's top 3. But individuals often move on to more personalized faire. So too with gaming. Tetris will always end up in the top 10 greatest of all time. It is a very safe choice. But most puzzle fans have other personal favorites (mine is Devil Dice). So people's favorites will often be less universally liked. For example, my favorite games usually get a 75%, which is the default score for a game that, while not ostensibly flawed, just didn't click with the reviewer. In a perfect world a game's potential for popularity won't affect the score but we all know what the reality is. You have to have your own filter when you are reading reviews, which is why I've had no problem buying games with low review scores.