Why Random is developing an irrational browsing experience


Sometimes a company is interesting not because of the utility of what it can provide now, but because of the slightly hard-to-pin-down possibilities it opens up — particularly if those possibilities may help us navigate a new age of user interfaces and experiences. Random is such a company, and here’s why.

Futureful iPadAbout a year and a half ago I wrote about a Finnish startup called Futureful for the first time. Backed by Skype co-founder Janus Friis, the company was building a StumbleUpon-ish content discovery app that also made website suggestions, but never gave the same suggestion twice, and always recommended a new range of keywords from which the user could choose her next adventure.

Futureful screenshot 3Last August Futureful resurfaced with a totally new design for its quasi-browser iOS(s aapl) app, based on a dreamlike, almost gamified system of keyword bubbles that one could combine and pull apart to find new content. And now the company has done it again – not merely redesigning the whole user interface but also adopting a new, easier-to-remember name: Random.

Outside the bubble

Gone are the green bubbles on the purple background, replaced by a more primary-color-based interface of keyword blocks that encourage you to go deeper and deeper through the suggestion layers. Touch one to bring up a page, then touch the color at the top of the screen to go back to the options. To back up through suggestion layers, just keep clicking on the white block in the middle of the keywords, which will take you further to the periphery of your demonstrated fields of interest.

Random screenshot 2What’s fascinating about Futureful/Random is that the startup doesn’t really need to worry about disorienting its users — as co-founder Jarno Koponen conceded, it doesn’t have that many. “We are still marginal,” he told me on Friday. The new app only relaunched in its Random incarnation a week ago. “The numbers of the past are meaningless to me,” Koponen added.

Random is in a good place right now. Friis’s continued funding means the small team doesn’t need to worry about monetization yet, and it’s pretty much free to keep reinventing its user interface in radical ways, learning along the way. But this is not an aimless journey, as Koponen explained:

“Our big dream is to create a new interface for the information around us. It is adaptive, it is predictive and associative. We have this big vision that there will be this new layer that is supported by a sophisticated AI that learns you. It won’t just learn rational you – it will also learn irrational you. We are not rational beings when interacting with any kind of content.

“What is the next UX when it comes to presenting information and enjoying content — what is the feedback loop? These are patterns that have nothing to do with rationality.”

Ever more explicitly, deep learning lies at the core of what the team is doing. The reason the clusters of keyword bubbles are gone, Koponen said, is that there’s no longer any need to show the user this information. “The artificial intelligence figures it out for you, based on what you do and what similar and current users are doing,” he said, explaining that removing the bubbles also takes away the conscious “cognitive load” from the user experience. It is now seemingly more serendipitous than ever.

Digging through the noise

Other people’s usage affecting what you see in the app, is a difficult factor to keep in balance.

Random screenshot 3When Futureful started, it felt like its content was too heavily skewed toward tech and design, because those were the big interests of its first adopters. Koponen insisted that Random is better at surfacing a wider variety of content, because it is more adept at understanding users’ intentions and interests.

He also said Random’s learning capabilities were able to develop recognition of when the user is picking a keyword that fits into her “long arc” of interest, or when she is just interested in something that’s timely or of short-term interest and unconnected with the long arc.

This is again where Random’s lack of monetization drive comes in handy. The team seems to be genuinely trying to see how keywords work when you’re not trying to make money off them, à la Google(s goog) or Facebook(s fb). Part of that means having no social layer, Koponen said:

“When you are trying to build a system that tries to understand you as an individual, the social layer can be too noisy. By [keeping] away from social, we’re also trying to democratise the data. Every keyword is equal. There are no categories in our system, just personal categories… an associative mind map.

“The thing is now we can try to understand how different individuals view different words. We are not tied to any ontology — there can be multiple ontologies.”

So, for example, if the user clicks on the keyword “Apple” the system can return suitable articles based on whether the user is already demonstrably a tech or a fruit fan. “We haven’t told the system what the apple means in this specific context, but the system figures it out for itself,” Koponen said.

Incidentally, Random doesn’t require any login via a social account – its use is completely anonymous.

Random experiment

So should you try out Random for yourself? I’d recommend it, though as one of the few users of the previous version it did take me a while to readjust. It’s certainly a lot prettier than it was before – a bit less playful, a bit more futuristic.

Random screenshot 4In fact, I would say the new user experience lends itself to futuristic forms of interaction, particularly non-touch gestures. It gives a much more literal feeling of what the system is actually doing – creating and digging through layers of association – and it’s a much more lean-back experience than Futureful was. An evolution of this idea could be great for virtual or augmented reality.

Like Futureful, Random is a great time-killer in its own right, once you’ve come to grips with the concept. But the best part of using it is knowing that you’re helping an effort that may be hugely profitable down the line, but is for now just interested in learning how people tick, with no apparent ulterior motive. Right now this is a tech rather than commerce play, and I’m looking forward to seeing where it goes next.


Tudor Lapusan

seems to be pretty nice this app.
Is there somehow a web/android version of it ?


Oliver and dotpeople, many thanks for your feedback (and interesting reference)! Keep it coming. :)

Tudor, thanks for reaching out. Android version planned, the schedule not confirmed yet. I hope you stay tuned.

-Jarno M./Random


“So, for example, if the user clicks on the keyword “Apple” the system can return suitable articles based on whether the user is already demonstrably a tech or a fruit fan. “We haven’t told the system what the apple means in this specific context, but the system figures it out”

This is reminiscent of E-Prime’s distinction between “observer” and “observed”.


“Korzybski (1879–1950) had determined that two forms of the verb ‘to be’—the ‘is’ of identity and the ‘is’ of predication—had structural problems. For example, the sentence “The coat is red” has no observer, the sentence “We see the coat as red” (where “we” indicates observers) appears more specific in context as regards light waves and colour as determined by modern science, that is, colour results from a reaction in the human brain.”



Hands down best way of finding new and interesting things to read! I got so bored reading always from the same sources. Random opened up a new universe for me.

Comments are closed.