Smart compose

I am recently relying more on smart compose when writing my emails inside Gmail. You get a suggestion for the rest of the sentence and if you hit tab it just gets written.

I wonder if Google will open it up making it available to 3rd parties. This will lead to explosion in writers. I wonder what if it ingests my blog posts and figures out my writing style, then suggests me sentences as I am composing posts. This will be interesting.

Disqus Reactions

Disqus – the commenting system I am using on this blog – recently added reactions, a quick way to react to posts without having to write a comment. If you scroll to the end of the post you will find the reactions buttons. They look like this.

What I like about this feature is the ability to use it effortlessly. Unlike commenting, you don’t have to login to Disqus to give a reaction on a post. You hit a button and it is done.

What is missing is dopaminification. As the blog owner I don’t get any notifications when someone react to one of my posts. I don’t even get a digest with the total number of reactions I am getting. It is also not clear which reaction is getting more clicked, which feels unnatural to what users are used to on FB. Sometimes familiarity trumps novelty.

This makes the reward cycle of this feature broken. People react to posts and I don’t get the dopamine rush of such acts, which should trigger me to post more, to get more reactions, and maybe on the longer run upgrade my Disqus account. 

Email reactions

Today I replied to an email with a thumbs up emoji. I just wanted to tell the sender that I got his message.

I am sure this use case happens multiple times a day. I want to tell someone that I noted their message (or love it, hate it…etc).

It would be cool if Gmail supports reacting to emails. Almost 100% of my daily emails come from coworkers who are also on Google suite.

The biggest challenge for this feature is 3rd party mail clients. Google can’t support this feature on there.

But they can at least add one click reaction icons. When the user clicks one of them like a thumbs up, it replies to the sender with an email with a thumbs up inside. Single click, works everywhere.

Blocking on YouTube

I don’t like Hussein Elgasmi. I blocked his channel on YouTube sometime ago thinking I won’t get his videos again. I was wrong.

As you see his channel is blocked, however I still got his latest video in my YouTube recommendations.

I am still getting Boshret Kheir in my autogenerated playlists. I always delete it when it is there.

It seems that blocking a channel and removing certain video/artist from your playlists aren’t strong enough signals for the YouTube recommendation algorithms.

I thought the former is a hard rule, not optional parameter for the algorithm. Maybe it is a bug, or no one thought of it.

Walking Speed

During Easter I traveled with a group of friends. Every time we wanted to go some place, we had to make a decision between walking and taking transportation.

I use Google maps, my friends use HERE maps. HERE always shows double the ETA of Google maps when it comes to walking. We kept arguing which is more accurate with no result.

Personally I have the same problem, my speed is mostly slower than what Google tells me. I thought Google does this based on historical walking speed for every person (I have location history enabled). Apparently it is hard coded at 12 minutes per kilometer, or more accurately, 20 minutes per mile.

We are afraid of AI taking over humanity, while someone at Google maps hard coded that all humans walk at 20 minutes per mile.

LinkedIn Closed Networker

People who add/accept anyone on LinkedIn (LION) to increase their reach might be doing it wrong. The way feed algorithms work – in principle – is they show your post to few of your connections, if they interact with it, the algorithm boosts it to more connections, and so on until it becomes viral, dies, or get replaced by other more viral content.

Unless your content is inherently viral, a big factor of people interacting with your content is them knowing you. Having a big network where you don’t really know the majority decreases the chance of your posts getting to someone who actually knows you in the initial phase, hence decreasing chances of interaction, which leads to lower virality and eventually less reach.

I am curious to see if LinkedIn data matches my argument (people with more connections, get less interaction on their posts).

The perfect echo chamber

I used to criticize Facebook for creating an echo chamber. My argument was – as almost every one else – that Facebook should show users what is right, not what they agree with.

I overestimated the level of transparency the internet brings. When the political events were happening in Egypt and the local media was hiding the truth, I thought as soon as every one joins Facebook/twitter the truth will be revealed. I was wrong. It happened to some extent, but the echo chambers were much stronger.

One of the main drivers of hostility on the internet is people seeing their core beliefs being attacked. Regardless of the side, seeing something we disagree with triggers our survival response and hence we become hostile to the adversary.

I am recently giving this a lot of thought. I started to think Facebook shouldn’t try to avoid echo chambers, it should strive to create the perfect one.

If you have the perfect echo chamber and only see the things you agree with, you won’t feel the internet is unsafe as it is now. It sounds counterintuitive, I know, but this behavior is why people are moving more towards private conversations.

That being said, it makes me question why we don’t have this until now, this is what I could think of

  • Nobody thought of it. I highly doubt.
  • It is not technically feasible. I also doubt to some extent.
  • It drives engagement down. If we only see things we agree with, we are less likely to engage with the content. Less engagement means less time spent on site, less ads to be served, and less money to be made.

That’s my theory.

Loading of Instagram vs Snapchat stories

I happened to be on a slow connection wanting to watch the current Snapchat and Instagram stories. 

Snapchat won’t let you open someone’s snaps before the app loads a few of them to avoid interruptions. Instagram will open the first story after loading the initial frame which sometimes open to you a frozen, waiting to load story.

Snapchat will start loading the rest of the snaps of the account while you are watching the first few. I think also Instagram does this. One key difference is that Snapchat won’t play a snap before it is fully loaded. Instagram does this like YouTube videos, the video loads while you are watching it (buffering). 

The problem with the Instagram way sometimes things freeze on the slow connection forcing you to close the story and reopen it. When you reopen it, Instagram considers you already watched the story you were loading and skips to the next one. 

This won’t happen on Snapchat because the snap won’t start playing before it is fully loaded. So even if Snapchat couldn’t buffer and you had to reopen the story, it will play the last one which you still didn’t watch. Unlike Instagram.

These small differences make me like the Snapchat experience more. I don’t have to take any further steps to ensure I watched the whole thing.

The developed world take some things for granted like a minimum connection speed. This is something I realized when I started working in Europe. Even when you simulate or run slow connection tests, most of the times you can’t get to a TE Data slow connection.

How Facebook recommends this random person you only met once

I wrote before about designing for creepiness, which is this simple idea that companies should inform users about why something is recommended to avoid freaking them out on how the service got to this info.

Recently in multiple conversations with people, it came to my attention that many are surprised by how Facebook recommended this person whom they only met once with nothing common.

For those who don’t know how it works, it is very simple. Facebook matches information from multiple sources to identify a person to recommend, then the software ranks those people based on a specific criteria that’s only known  to those inside Facebook, and decides which accounts to display.

So, here is a list of possible ways by which Facebook recommends friends (I don’t know which ones they actually use):

  • If you give Facebook access to your email contacts, it will match those emails with Facebook accounts and starts recommending users using Facebook with the same email in your contacts list.
  • If someone has just joined the network and started to get a lot of friends who are mutual with you, Facebook will assume you also know this person and will recommend him/her for you.
  • If someone is using the same computer to login to Facebook, Facebook might recommend friends of that user to you on the basis they might be mutual. (If you use multiple accounts on the same computer you will notice this).
  • Facebook owns WhatsApp, for WhatsApp to work it needs access to your phone contacts, and it uploads them to the servers. If Facebook knows that I have your phone number on my phone it can start recommending you to me and visa-versa.
  • If you and I meet in the same location everyday from 9-5 (Facebook knows that from our phone location, or the network we use to login to our accounts like work network). If one of us lists on their account where they work (Or even without that), it will consider us working at the same place and start recommending us to each other along with other work mates.
  • If we both attended the same event (Visited the event page, said attending or maybe..etc) then our GPS locations confirmed our location at the time of the event, it might start recommending us to each other.
  • If we are in a group picture with mutual friends, it will definitely recommend us to each other.
  • This one is interesting: If you stalk someone by visiting their profile, Facebook can start recommending you to this person.
  • If two friends have your number/email, one of them is your friend, Facebook will recommend the other one as your friend.

This is what I got on top of my mind, what other ways Facebook might be matching information to recommend you friends?

Signifiers for tweetstorms

A key design element to any object whether physical or digital is signifiers. Signifiers tell people how to interact with the object, unleashing its capabilities.

Affordances define what actions are possible. Signifiers specify how people discover those possibilities: signifiers are signs, perceptible signals of what can be done. – Don Norman, The design of everyday things.

It is surprising, how a company like twitter suffering from an engagement problem partly because of the 140 characters limit, yet users started using the product in a different way to overcome this problem by posting tweetstorms.

Tweetstorm is a series of tweets about the same topic, posted sequentially and mostly marked by their order. – Unknown

While twitter now supports tweetstorms by allowing users replying to themselves, there are no signifiers on the app nor the website that you can do this. I still see lots of tweetstorms where users post each tweet separately marked by a number instead of replying to the previous tweet to continue on the topic.

This problem leads to another problem which is making it harder for users to read the whole tweetstorm specially if it is from a user they are not following, because they can’t simply tap on the tweet and read the whole tweetstorm, instead they have to visit the person’s profile to read everything. Add to this if the tweetstorm is old and unlinked, it is very hard to reshare it whether on twitter or on any other platform as the users have to scroll back in the person’s profile until the time of the tweet to be able to read the storm.

Twitter, please fix this.