Thanks! That’s actually pretty darn close to what I want to accomplish. It really helps 
1 Like
I dislike the approach (which both Inform and TADS run as well) because nobody does it. Start and end the transcript, search for the file, fire up the email program, search the game for the author’s address, copy it to the email program, attach the transcript - There’s so many reviews and ratings over at IFDB.org, I wonder how few players actually sent in a transcript. My guess is 0.1%.
1 Like
Just to wrap this up for me, my preferred approaches would be:
Okay: Option for the author to set a flag that asks the player at the beginning of the game: “Would you be willing to send a transcript of your game session to the author so that they can improve their game? (Y/N) (Your default email program would be used and the author can see your email address.)”
Perfect: Same opt-in question but automatic saving of transcript to my e.g. Itch.io account. No idea whether that’s technically possible. Iirc Bob Bates did something like that when Thaumistry was still in beta stage.
On the one hand, I think the standard emailing-the-data approach is a fine and trustworthy way of doing things, and a perfectly valid design choice, but on the other hand, I tend to agree with Grueslayer that this basically amounts to forgoing the opportunities of telemetry.
I mean, the attractive point of telemetry is that it ideally is a frictionless, streamlined, as-automatic-as-possible way to gather data (while still requiring consent, of course!), with minimum effort for players and authors, so that hopefully authors will be able to glean more insights and have more evidence of how their games are played.
How about something like this, coming up when the player is quitting the game or has completed it:
It’s just a quick mockup, but I think it shows the idea:
- Display the actual data that would be sent. (And unlike OS/software error logs, it would mostly be comprehensible to a layperson.
) - The “No, thanks” button is the default button.
- The option to save the log gives the player time to think it over.
- Player needs to check the checkbox to give explicit consent.
In case of consent, the data would be transmitted via HTTP.
3 Likes
As a player, I’d certainly be a lot happier with that than with just an opt-in at the start that I might miss or not think about too much if I was keen to get on with the game (a notification at the start and prompt to give consent to send at the finish would be ideal).
1 Like
I like this more refined approach even more.
There’s a technical challenge with this in a web context, since a tester/player is free to shut down a browser tab or a browser window at any time, destroying whatever gameplay data has been stored in the browser.
I can think of one solution for this right off the bat, but I’m sure there are many. For savegames, trombicula is currently utilizing “Web Storage” (localStorage). To avoid losing telemetry, the “audit trail” could be stored automatically in a similar way (if enabled). The moral quandary of this, is that the interpreter would be writing data to the tester’s/player’s computer on its own volition.
1 Like
Yes, that’s a difficulty… not sure how one would best address that.
It could be done via an additional opt-in at the start, but that could have the effect of scaring off some players, even though it’s technically innocuous. In effect, it’s roughly the equivalent of the scrollback which other interpreters keep anyway (plus maybe some diagnostic data), except that because of the reason you described, we can’t easily go the way of just keeping it in the RAM and offering to save it on closing.
Sorry for the double post, but I wanted to add two aspects which merit this separate post IMHO:
(tl,dr: HTTP telemetry is more anonymous and induces less of an obligation to write additional comments than e-mail.)
First aspect
When e-mail is the only route on offer, you’re requiring the player to give up anonymity, more or less. (Well, e-mail addresses don’t need to contain the real name, of course, but you know what I mean.) Now the transcript and all its commands and comments are directly associated with a specific mail address, and thus, with a specific person or online persona.
Whereas with a transcript that’s directly transmitted via HTTP (with explicit consent), the transcript is only associated with an IP address in the server logs, which is in itself not a very interesting or compromising datum about the sender, except in a law enforcement context.
So, in addition to the lower-friction and ease-of-use aspects, maybe some people would rather send their transcripts over HTTP than via their e-mail address, when they contain stuff they might consider embarrassing (failed at an easy puzzle, or tried to >BURN ORPHANAGE, or tried to >HUG and >KISS THE EVIL OGRE, or whatever ).
Second aspect
When players are required to send an e-mail, it might be the case that they feel obliged to write some additional comments in the e-mail, which they then put off to another moment, and then it never happens.
Of course, it’s not much of an effort to write “Hey there, I enjoyed your game. A transcript is attached. Regards”. But I know that I myself have occasionally put off writing such a mail, because I felt that the game merited more words from my part, and it would be polite and appropriate to give the author more feedback.
It’s a bit irrational, because one knows that for the author, even a transcript without accompanying letter is much better than nothing… but people are irrational like that sometimes (myself included).
So, that might also be a point in favour of adding telemetry, where players wouldn’t feel such an obligation, and would just hit the “send” button on the provided form.
1 Like
Quest 5 runs on the server; all commands go to the server, it parsers them, etc, then returns a response. It is already “phoning home with details of a play session”. Is that creepy?
I am just putting this forward as another aspect to be aware of. Quest 5 could log all player interactions, but does not; from a practical point of view it is no different to playing in your browser. To be honest, I get that it is creepy recording what people do without them knowing.
However, a bigger consideration for me is the sheer amount of data that would be produced. I have a game being tested at the moment, and just the transcripts e-mailed from one (very thorough) tester takes a long time to go through.
[Quest 6, by the way, runs in the browser; game files are downloaded and that is the only interaction with the server.]
1 Like
My own point of view is that some friction is how you are sure someone is REALLY consenting.
When sending out to beta testers, you instruct them how to take transcripts. This means you forgo a lot of in game telemetry for normal players but I’d rather have less useful data than do something the player is not really aware of or creepy.
1 Like
Totally agree with this. How much of a difference is there between a “beta tester” and a “player” in this regard though?
What you propose is likely to be legally compliant, and some at least would probably be happy to provide the requested information.
I have to admit that I have bailed out of games that had such a prompt during/at the end of gameplay before. It feels like the game is watching over my shoulder in a way that breaks immersion, and I worry about it being insecure because of the bad example of the likes of Google and Facebook (where they used similar styles of consent message to some things as cover for illegally taking data in other ways). Finally, I’m accustomed to most games being possible to play off-line; a “phone home” feature, even if unused, isn’t always compatible with that preference, depending on how connection consent has been done. (Obviously, the last obstacle isn’t a concern if the game was online or relies on an online component for something obvious like copy protection or synchronising achievements).
The automation of telemetry is, itself, the problem - I also don’t like the telemetry from Steam or GoG Galaxy, even though as far as I know both of their (partial) telemetry systems have been implemented as benignly as could reasonably be expected. I agree with @adventuron that without friction, there can be no surety of consent.
Ren’Py (the platform I use) saves all data to save slots, including any telemetry the author may have chosen to implement. Saving data to the player’s own computer isn’t the problem - it’s letting the data be anywhere other than the player’s control. A player can delete a save on their machine, but can’t recall information that’s been sent to a server.
(I also object to sending anything that could possibly include personal information - include the character name, if the game had the option of renaming the character, as well as the machine I played it on and my preferred response method - via HTTP. It’s simply not a secure protocol these days. I’d expect HTTPS or something else secure to be in use, for an in-game sending method. (The good news is that isn’t much more complicated to implement for an author, and there’s no reason why this would result in more complication for a user. Ideally, there’d be an option to remove sections, either due to sensitivity or because it was being sent in a context where only part of the transcript was helpful e.g. you’re quitting because of an annoying bug in room 6).
2 Likes
I definitely get both sides of the argument here, but I have to say, as someone who sees a lot of value in providing transcripts to authors and tries to do that whenever possible when writing reviews, I’ve never actually emailed one to an author even when it’s requested in the ABOUT text (except of course when beta testing). Manually emailing something to someone you’ve never met isn’t just adding an extra step, it also can twinge folks’ social anxieties in ways that really discourage the practice. Things can also go awry with locally-saved transcripts – I’ve had mid-game crashes mess them up and accidentally overwritten them, and when I was testing Faeries of Haelstowne I even exceeded the maximum transcript length (admittedly it’s a very big game with a lot of words).
To give a personal example, I got maybe half a dozen emails from folks giving me feedback on my IFComp game last year, I believe only one sent a transcript along. So based on reviews and that transcript, I think I had a list of fixes for a post-Comp release that had about 8 or 9 items on it. But then I was able to look at like 50 additional transcripts from folks who played on the Comp site, and I wound up with over a hundred additional bugs, quality of life improvements, and other upgrades from that.
Of course, clear disclosure about what’s going on, conscientiousness about security and what’s going on on the back end, and an easy ability to opt-in or out or important, but it seems like there are people, and systems, currently set up to do that and that are working well. I’m actually curious if there’s been any feedback on how the Comp does things, since I believe the transcript-saving has been up and running for a while, and doesn’t have some of the features folks have discussed (the only opt-out is to play offline, I believe).
6 Likes
Beta testers are testers that you can expect are happy to provide transcripts, and will not be creeped out even by the mention of instructions to enable them. Regular users are users of the live game which may be creeped out even by the suggestion of keylogging them.
The time for telemetry is in the beta testing phase.
Once a game goes live (non beta), you ask for telemetry at the risk of your reputation, and hide your telemetry at the risk of the law and your reputation.
4 Likes
It is, a bit, and also the reason (I believe) that the Quest 5 online playing experience is frequently so impossibly slow. Sounds like Quest 6 is a definite improvement!
I believe you can add “?feedback=0” to the url of the game to opt out.
2 Likes
Yes, hmm, I’m kind of torn on this myself. I understand and partly share your and adventuron’s point of view and concerns, especially about companies’ shady practices of data collection, profiling, and tracking.
I think with my proposal, I was (and am) basing that on a view of the IF community and indie game dev scene as being more benign in their (our) practices. But I can understand if we don’t want to foster a maybe false sense of security in our users, which would do them ill to transfer to the context of the internet at large.
About the feeling of being watched by the game: Yes, I can relate to that.
The way I was envisioning it, it’s less as if the game is stealthily gathering all sorts of info in the background, but more intended to be like the game saying: “Hey there, thanks for playing. If you’d like to, you could do CTRL+A, CTRL+C in the game window and paste that into an email. But look, I already copied the scrollback buffer, which I was keeping anyway, into this textbox for you. Would you like to send that?”
(That’s not meant as a counterargument, it’s just to clarify the perspective I was taking.)
Regarding offline play: agreed, and if some sort of telemetry was implemented in an interpreter/platform, it would always have to degrade gracefully if the player is offline, and should not interfere with the ability to play the game. (For example, the interpreter could show a window similar to the one pictured above, but just offering the option to save locally.)
About HTTPS: yes! 
I was just using HTTP to refer to that protocol in general, in contrast to e-mail or other channels.
3 Likes
I avoid online play for any type of game. Telemetry, tracking and privacy are part of the reason.
I constantly have to prune and block unwanted emails, spam and phishing.
5 Likes
I second that. If I’m beta-testing a game, then I’d be happy to email or send whatever transcript needed to debug the game. If I’m playing, then I don’t want to have telemetry.
“Select all - Copy” isn’t that difficult. If the author wants it emailed or uploaded somewhere on his cloud, then it’s fine by me.
1 Like
Select-all doesn’t work. Interpreters usually trim scrollback after a certain point, and will certainly do that if the game has a “clear the screen” command.
2 Likes