Writes Like A…
Whilst I love Eurogamer, the site’s just too cluttered for me to read on a regular basis. Whenever I do go over there for a review or otherwise, it’s always the comment threads that leave me most entertained. My favourite kind of comment is one that seems more prevalent at EG than anywhere else.
“Good review. Reads like a 7 though.”
So I went to find just how prevalent these things are. Cue small amount of programming…
I wrote a tiny web script that would run over every reviewer on the EG staff, and look at their most recent 300 reviews. It grabbed every comment and did some (very rough) guesswork as to how often these things come up. Here are some NUMBERS labelled with WORDS representing what I found:
There’s a whole load of ones with two or fewer – in fact, there’s a “Reads like a…” comment for every ten reviews on Eurogamer, which is a fairly high proportion. And that ignores all the ones I’ll have missed out because they used wording my quite simple script didn’t pick up.
Here’s some details on the writers too. Our top five, by total number of disputes:
And sorted again, the top five by disputes per review:
Some Words On The Numbers
First of all, the data is really shaky. I didn’t do this very rigorously, and anyone looking at my source code will be able to point out a ton of mistakes made in gathering the data. But it’s here as a guideline, just to look at and perhaps provoke further looks by more dedicated people.
Secondly, this isn’t meant to be a reflection on those writers named above. In terms of the top five by total disputes, two of them have over 300 reviews (making them long-standing writers at EG) and another two – Ellie Gibson and Simon Parkin – are reviewers that people love to discuss and argue with. Naturally, these people are going to garner more of these comments than others. In Christian Donlan’s case, he was just in the wrong game at the wrong time – the game which ended up topping the list of disputes-per-game, Mirror’s Edge.
The type of comment I searched for irritates me – and I would think others – because it has no real purpose. Ignoring for a second the fact that scoring systems are a dubious way of estimating a game’s worth, the idea that a bystander would be able to better make that conversion from the words of the review to the numbers of the score is absurd. When I used to write the odd review for Pocket Gamer, the difference between a 6 and a 7 could come down to which side of the bed I woke up that morning, or whether I liked the colours in the menu screen. The review is the opinion. The score is the shoeboxing, the categorisation. It’s an afterthought, a bad habit we can’t drop.
That’s a discussion for a more verbose time, I feel. I hope the numbers were vaguely interesting. The raw data is available here.
The Code Behind The Numbers And Words
So I finally found a Python HTML scraping library that really works. If you’re looking for one too, try BeautifulSoup. It’s fantastic. My source code can be found here; it’s messy, but it’s there. Running it on a MacBook with a reasonable connection, it still took a couple of hours to chew through all those reviews. Feel free to modify it and run it yourself.