I was interested in snaphat over the weekend because it updated and changed how information was presented in a very major way that was surprising to me. The first thing I noticed was when updating the app it doesn’t have a good rating, which doesn’t cause us to act in any particular way, I just thought it was something really strange because it’s so popular and is even #1 in photo and video. Why do we all use it if we don’t appear to like it?

Onto the update functions I noticed, the first change I noticed was the screen to select who you want to send a snap to. It is broken into more clearly separated groups of people which are stories, best friends, recents, groups, and friends. This constructs the user by making it easier to send snaps to certain people. The way it’s organized makes it so you want to post a story or send it to the people you are already sending messages to because those are the top categories. It isn’t much different than before other than it looks nicer, but the group button was what really caught my attention. Because I don’t have any groups on snapchat I was presented with the message “You don’t have any groups yet! 🙃” The upside down smiley face always reads as really passive aggressive to me, so this constructs the user in a way that tries to make you feel bad for not using this function. It’s saying “oh you don’t have groups yet. ok it’s fine it’s whatever even though you should totally do it i’m not mad about it it’s cool your loss.”

Finally, the biggest change and the one I’m least happy with is in the way stories are presented. I follow Cr1tikal on snapchat who is a pretty popular youtuber, and he posted a story on snapchat. Later in the day when I got the update, I noticed that in the section where stories used to be, his looked like one of the weird articles that snapchat tries to push it’s users into reading. None of my other friends stories were there, his was only there because he is a popular person. It even says popular story on his image. I kept scrolling down to find people I knew in this section, only to see a bunch of random people’s stories or articles about something or I don’t even know what. Like who is the user “Chunky?” I don’t know her. I don’t want to see her story. To find my friend’s stories, they were moved to the list of friends. If you swipe left, they are now there and if you swipe right is where all the articles and random people are located. This constructs users by separating who you know into a private location and eveyone else into another location, which I guess makes it more organized? But it’s just confusing and unnecessary. I think it’s also pushing the idea that you should care about these people. They’re popular, so you should care. Your friends lives don’t matter as much and their stories are presented in a less obvious way now because of that.

When scrolling through instagram I noticed this “Popular Hashtags” post placed in between the posts of people that I actually follow. One thing that I think is interesting about this project is that I am paying more attention to everything that pops up on every piece of social media that I use, and I don’t think I’ve ever noticed this popular hashtag post in my feed before. Next, the phrasing of it itself stands out to me because it says that these searches are popular, and although they might be popular for a very specific audience, it’s clearly just searches directed at me rather than what is actually popular. For example, #nofilter has 216,267,715 posts and what I am being shown (#igbulldogs) has 588,442 posts. If I was really being recommended hashtags that are popular I would be seeing completely different images. This shows how instagram is keeping track of who I follow and what I search, like other sites and advertisements. Finally the follow button catches my attention because I have never followed a hashtag before. Instagram has recommended users show up all the time that they want me to follow, but I’ve never seen recommended hashtags to follow. I don’t even know what this would do, would I just see every single post that shows up in the tag on my feed? Why would I want to do that?

I focused on the website Quick Draw, which is a game from google where the user is given a word to draw in 20 seconds and an AI will try and guess what is being drawn. The user is providing drawings, which gives data to this AI so that it can become smarter and learn to recognize and distinguish different things from one another. It’s typical use to be played as a fun past time and the audience is likely anyone because it was used for data collection, but I think it was mostly used by younger people since it became popular through social media like Tumblr and YouTube.

Day 1: When I started on day 1, I just had the idea that I wanted to draw what I wasn’t being told to draw. My original idea was to see my art among all of the other drawings and if other people didn’t draw what they were supposed to or how long it would be before it got taken down. After my first round of drawing, I went through the data to try and find one of my drawings and I noticed that it wasn’t organized in any particular order, and the only dates I saw were in January or March of 2017 (which maybe was just coincidence with all the random ones I clicked but I clicked a LOT.) Knowing this, I assumed my drawings weren’t shown because either all of the data wasn’t shown or it was so randomly organized I never was able to find them, but I think that’s more unlikely since all of the months were only between January and March. Instead of observing my drawings in the data sets, I changed my plan to be to test the game and the AI’s intelligence by drawing what was asked in less traditional ways. I also tried to do my best and “win” by drawing what was asked, but I was never able to win since I drew differently than how the game is normally played. For each day I played the game about 10 times.

Day 1 Example: I didn’t know what I was doing so I drew cats at first since that was the example I gave in class. Halfway through, I thought I could draw cats vaguely shaped like what was being asked, as shown by the results above.

Day 2: I noticed when looking at the data sets that most drawings were drawn in one particular angle, so to test the AI’s knowledge I would draw everything upside down. Most of my drawings weren’t guessed, but when they were I noticed that the drawing could be reversible so they would make sense upside down. To do this, I started by drawing in a sketchbook to have reference and then transferred that to the computer so that it would be easier for me and more accurate for the game to guess.

For example, the AI wasn’t able to guess snail, which to me looks pretty obvious. It was able to guess toothpaste since all of the drawings of toothpaste were drawn from all sorts of angles, and no angle would technically be considered upside down. 

This one surprised me because I feel like a sword can be clearly distinguishable from all angles, but once again all of the data collected is drawn one way so this could lead my drawing to be taken out of the picture. Out of all the drawing tests I did on the program, doing them upside down was the most fun for me. None of my drawings are wrong because they’re upside down, so if it could be found in the data there isn’t a reason anyone should report it and it’s teaching the AI to see things in more ways than one (you know, maybe, or maybe not since I’m just one person so the drawings might not do much.)

Some problems that I think could stop this test from giving the best results are drawing ability, the times given, and how the program has a limited list of things to guess from which might make it guess the right thing because there’s nothing else left.

Day 3: For day 3 I did 2 different drawing tests. The first one was based on the fact that I noticed how little the program would need to guess certain things, so I wanted to draw the words in the most minimal way possible; by using only basic shapes. This was the closest I got to getting 6/6 right out of all of my attempts, which I think shows that the AI can recognize things based on position once again and simple structure. I think what ended up confusing the AI during this the most was the extra lines and spaces since I was trying to separate the shapes so what I was doing would come across more clearly, but it still was able to guess a lot right.

This was a good example to me of how little the AI needs to recognize things. I was really surprised at a lot of the drawings it got right, and I think looking at the data from these tests probably shows the most interesting results since it is so surprising in some cases.  

It was also really funny to me because I ended up drawing really similar stuff a lot of the time because I wanted to keep it simple. The ones guessed right that I drew basically the same you can kinda tell where they were coming from, but they really don’t have that much in common at all.

Some problems with this test were that some of the words were shapes, so drawing that made it easier. I also found it hard to draw the most simple form of something and definitely think that I could have drawn too much which may have skewed the answer as well. Finally I also think that the limited list is a problem once again, and I wanted to explore that idea as well.

Extra test: I wanted to try and draw other words that appeared similar to the word being suggested. The game tells you what the AI thinks you are doing before it guesses correctly, so I wanted to try and take those original guesses and draw those instead. This test was the least interesting because I was drawing a different word completely than the one I was supposed to, so the AI always / almost always guessed wrong. It did however guess the word that I was drawing each time so it did know what I was actually drawing.

What caught my attention this week was the addition of gifs for stories on Instagram and how people have been using them. This feature came out this week and when I read about it, I thought it was pretty dumb but this is likely due to the fact that I have never used Instagram stories so this feature would be useless to me. What I noticed from how people I follow have been using this new function is that there has been a trend where people “post their aesthetic in gifs,” or another one where you “search your name and the first gif describes you,” or things of that nature. I haven’t really seen anyone use gifs on Instagram in another way yet and I’m curious as to if this is just the people I follow or if this trend is on most accounts right now. Another thing that I think is interesting is that Snapchat added “animated stickers and filters” last year. Instagram has taken a lot of ideas from Snapchat, like stories for example, and the timing of animated stickers and gifs makes me wonder whether it was a coincidence or not.