Digital Clones, Alternative ISPs, Closed Networks and Brainwave Tech
Hi everybody, thanks for reading the second issue of our newsletter. This month we’ll talk about brainwave headsets, digital clones and private networks. Hope you’ll enjoy it!
Your brain is a semi-open book: closing your eyes and your mouth is not enough to keep your secrets.
Let’s face it, in 2018 we are still just scraping the surface of how the human brain works. The best place to start, if you have a full day in front of you, is to read Tim Urban’s post.
In terms of products and companies, CBinsight and Crunchbase published two excellent reports about VC investments in Neurotech which are on the rise ($600M was invested in the recent months). If look at the list of companies working on Neurotech, you will notice that the majority is based on ElectroEncephalogram.
Electroencephalogram is not a new technique. It has been around for close to a century (the first EEG of a human brain was recorded in 1924), and it consists in “reading” the brain electric activity, a.k.a the brainwaves, thanks to electrodes placed along the scalp.
I am not going to rewrite wikipedia but here is a simple definition I found:
“Brainwaves, also known as neural oscillations, are the rhythmic electrical activation of large groups of neurons. These oscillations are labelled according to the frequency with which they occur, and different frequencies are loosely correlated with certain kinds of cognitive function.” Source
To keep it simple you have electricity going through your neurons when they communicate information and seen from the skull this creates an electric wave with different frequencies and each frequency will indicate a certain level of focus or relaxation.
Keep in mind that EEG is a non invasive technique that catches the signal from outside the brain. This is where software becomes useful to first clean the signal and then interpret it (here is a good Quora post which explains how ML is used for that purpose).
What is interesting is that this technology was mainly used and developed by researchers and doctors in labs and hospitals, but now more and more entrepreneurs are working on brainwave products targeted directly at the end-user for very specific use cases that go beyond healthcare.
Basically we’re going from this type of headsets:
To this type of headsets:
What is noticeable is that these startups focus on very specific use cases like losing weight, improving the way you study or monitoring your sleep quality and many more. They can’t compete with multi-purpose / high end headsets in terms of performance and precision, so they concentrate on “narrow applications”.
Clearly we’re only at the beginning of this trend, a lot of readers will probably tell us that these startups are “over-promising” and that the results delivered by these headsets are probably limited at the moment (yes, some are probably “bordeline” scam).
We won’t argue with that. We are not experts and have only tried some of them over a short period of time. However what’s interesting is the dynamic of the trend and to think of the bigger picture (after all it’s our job :-)):
- We expect an increasing number of “brainwave — EEG based” tech startups to emerge in the years to come.
- They will address an increasing number of use cases not necessarily linked to healthcare. Ex: a headset that you use to monitor and improve your performances while playing video games, or that you will use with dating apps to visualize how your brain reacts to the pictures of a potential date and see if both users are compatible).
- The software and hardware components keep improving so it’s not crazy to think that these dedicated headsets will slowly disappear and will be integrated directly in your glasses or in the headphones you use to listen to music (and will make recommendations based on how your brain react to the music you listen). Think of Fitbit trackers which started as dedicated fitness wristbands and are now integrated in watches, rings or earphones.
As a conclusion it’s worth insisting that EEG is only one of many methods that let one “read” the brain activity. Several other approaches exist such as fMRI or electrodes implanted directly within the brain. We’ll try to cover some of them in the coming months.
Our must-reads this month:
- The brain is a predictive machine. Here is a fascinating article that explains how the brain is an anticipation machine and adapts with experience.
- Computer synapse : we talked a lot about the brain in this month post, but computer chips are starting to mimic brain synapses.
- A good piece in the Guardian on a journalist testing different cognitive enhancement tools and drugs.
- MindID: Person Identification from Brain Waves through Attention-based
Recurrent Neural Network.
Space/Time & Reality
Meet your digital clone
You’ve probably already seen this video of a fake Barack Obama:
Tremendous progress is also being made on the voice side. It’s becoming almost impossible to tell the difference between an audio recording made by a human from its fake AI generated version:
So you probably see us coming, if you combine both you’ll soon be able to create fake videos of people saying stuff they’ve never said, and it’ll be impossible to tell whether it’s real or not.
And it’s only the beginning. We think that the next step will be to imitate/clone your personality. Feed a deep learning machine with all your emails conversations, your WhatsApp messages, your Facebook and Twitter feeds, the interactions you have with your voice assistant (Alexa, Google Home, Apple Home), the blog posts you wrote and, boom, an AI will be able to imitate some aspects of your personality and reproduce some of your habits.
Such technology could be used, as usual, for good reasons (as your personal assistant who’ll answer some of the messages you receive on Whatsapp or Facebook), for bad reasons (to steal someone’s identity) and for more “awkward” reasons (create the digital clone of someone who died so you can interact with her/him).
Another parallel trend will grow in response to these applications: services that will help you tell real from fake. Factmata is an ‘anti fake news’ media platform which recently raised $1M; We wouldn’t be surprised to see similar services emerge to help people spot AI generated videos, voice recordings or even “fake” online interactions (Is it really you who posted this message on Facebook?).
Our must-reads this month:
- Related to our discussion just above: “Molly wants to use your online presence to create an automated knowledge base” and “We used predictive keyboards trained on all seven books to ghostwrite this spellbinding new Harry Potter chapter”.
- Self-Awareness: the Last Frontier.
Surveillance & Privacy
The future of the internet might be Alternative Internet Providers & Closed Networks
Before we dig into the rise of alternative Internet Service Providers and closed networks it’s maybe worth recapping how the internet works (we’ll keep it short, don’t worry). Basically internet is a network of networks. It’s a huge number of networks which are interconnected.
But all networks are not equal in importance, there’s actually a hierarchy.
Let’s illustrate it with a concrete example. Let’s say you setup a home network, connecting your computers together through a WIFI, a LAN or whatever you want. Your network is quite limited (it’s shared across a couple of computers only), but it’s a network. Now, if you want to have access to Facebook, Google or other websites, you need to connect your home network to another network which is connected to Facebook & Google servers. And this is your internet provider (ISP).
The biggest internet service providers belong to “Tier 1 networks” (many are big Telco companies) and are on top of the food chain. They can access every public network available directly or indirectly (this is why you pay for them), and they own a big chunk of the internet infrastructure: from the cables and optic fibers installed in our cities to the huge undersea cables which connect the different continents. We won’t go into too much details, but other lower ranked Internet providers also exist (Tier 2, Tier 3). They own less infrastructure (sometimes even none) and need to “pay” their access to the internet from the bigger players. Your home network is, sorry for you, really low ranked in this hierarchy and doesn’t have much power.
This hierarchy explains why:
- Many Tier 1 networks are against Net Neutrality. Since they have a dominant position and own a big chunk of the pipes, they want to be able to charge differently depending on the source of the traffic they route.
- Many Tier 1 ISPs are not interested in serving smaller markets like rural areas.
- Companies like Google and Facebook want to own more infrastructure (E.g: Google optic fiber).
The rise of “alternative Internet Providers”
In this environment we’re seeing an increasing number of alternative, and very often local, Internet Providers being created. These smaller internet providers can be private companies, associations or even local administrations (community broadbands) which provide internet access to their communities with various aims in mind: granting access for underserved populations, for cost reasons, to guarantee Net Neutrality, to push for local economic or social development, to guarantee privacy and more.
For instance the number of community broadbands (internet provided by municipalities) in the US has almost doubled the last three years from 450 listed in 2015 to 750 in 2018.
Even in France, where current regulation ensures Net Neutrality, forces the major ISPs to cover rural regions and forbid them to monitor and resell their users’ consumption habits, several alternative ISPs exist (for example Illico an internet provider which is run by a local association preaching for Net Neutrality). Another great example of such providers is Gufi in Spain which has over 33,000 active nodes and about 46,000 km of wireless links.
But how can they provide internet independently? As I explained above “anyone” can start its own Internet Service Provider by renting/buying access from the bigger players. Some of these alternative providers are “purely virtual” (they own no infrastructure and only rent their access) while others build the local infrastructure (local governments installing fiber optic or associations mounting antennas on buildings for wireless internet) and they connect their local infra to the main networks directly at Internet Exchange Points (which are building where ISPs exchange Internet traffic between their networks).
We definitely think that we’ll see more and more of these alternative ISPs emerge and that an increasing number of users will choose them over traditional Telco Companies. See also initiatives such as SpaceX Starlink recently.
The rise of “closed networks”
Another very interesting trend to follow are “closed networks”. By closed networks we mean networks which are not connected to the internet. If like us you’re in your thirties / forties (;-) Nico), you’ve probably experienced closed networks when you were younger through “LAN parties”.
But you can do much more than playing games. This article on Wired describes the closed network that Cubans built because they were frustrated with the connexion provided by the government. The “real internet” is barely usable on the Caribbean Island so people have built a local wireless network accessible to anyone, not connected to the internet, where they can use messaging apps or download content from the local servers (people who are flying back to Cuba are bringing with them hard drives full of content that they upload on the local physical servers so that everyone can benefit from it).
What makes closed networks more and more interesting is that the technology to run them is becoming mature and the applications that can run on them more sophisticated.
For instance in terms of infrastructure mesh networks are getting increasingly popular and easier to use. If the dominant model on the internet is the server/client one (when you send a Whatsapp message to your neighbour, the message will first be sent to Facebook’s servers thousands of miles away and then it will be routed back to your neighbor), Mesh Networks have a more “peer-to-peer” approach. In such networks each person is a node that can transmit data. So when you send your message to your neighbour it’ll travel directly through the fastest “local” path. And there’s no central entity that can take the network down.
GoTenna is a good example. The startup sells a piece of hardware that you pair with your phone and once it’s done you are part of their Mesh Network. You can send and receive messages from other members without having an internet connexion (if you have other members close enough to you, otherwise there’s no magic you cannot reach anyone).
If the main use cases of Mesh Networks are messaging and location sharing, which are especially useful in the context of natural disasters and political upheavals, you can actually do much more than that.
Scuttlebutt is for example a social network that can run completely on “closed networks”. Services such as Pirate Bay showed that it was possible to share big files in a P2P way, so it’s possible to have the same model running on closed networks (and watch videos). We’ve also seen some B2B startups offering “closed network” products to businesses: the applications running on these closed networks have less chanced to be breached from the outside and can run faster.
We also believe this trend will spark a new wave of personal servers that people will keep at home (or in their pocket). It started a long time ago with people storing media files (music, photos and movies) on their home “NAS”, but what if people started to store apps or their personal data? It would be first stored locally, and then sent to external services (think of a segment.com but for your personal data).
It’s also worth noting that many of these closed networks, and the applications which run on them, can also work while connected to the internet. Think of your Google Docs that you can edit on your browser while offline and that will be updated online once you are connected, this model make sense for plenty of apps. Scuttlebut, the social network I spoke about above, can work 100% in “local mode”, but you can also setup a web server so that people on the other end of the world can read your posts. This “hybrid” mode has a huge potential.
Why are we speaking about these two trends in our “Surveillance & Privacy” section?
Because these models are a great counter-balance to a “centralized” internet where the power might be too concentrated in the hands of a few big internet companies and ISPs. What is currently happening in the US (Net Neutrality is threatened and ISPs can already sell their customers’ surfing data) can also happen in other parts or the world (several big telco companies are pushing for the same rules in Europe). People are increasingly aware of these problems and embrace these alternative models, whether as a complement or as standalone solutions.
What is also exciting is that the infrastructure needed and the software that run on these closed networks are only getting better. We’re still very early but there’s a lot of interesting stuff happening in crypto & blockchain tech that could be applied here. We expect a lot of interesting startups to emerge.
Our must-reads this month:
- The best patents of the month goes to: Amazon Patents Wristband to Track Hand Movements of Warehouse Employees & Facebook Patents Tech To Bucket Users Into Different Social Classes.
- Facial Recognition Is Only the Beginning: Here’s What to Expect Next in Biometrics on Your Phone.