MEX

Microsoft Research has suggested a method to overcome the latency which can hamper the user experience of network-hosted services, such as connected Xbox games. The DeLorean approach relies on predicting user behaviour so that servers can calculate several possible future outcomes concurrently. By combining these predicted outcomes with bandwidth compression, the server delivers multiple possible outcomes to the client in advance, thereby masking the latency incurred by network transport. The client device is then able to render whichever outcome occurs almost instantaneously.

Microsoft’s Research team focused on its application in Xbox gaming, but the technique would seem to be just as relevant to other forms of networked services, such as mobile apps. It could also help to mask the inherent unreliability of wireless coverage.

Analysing user behaviour to create shortcuts to future outcomes is a well established technique, underpinning everything from input systems like Swiftkey to services like Google Music, which store most of the user’s collection on the network, but cache frequently accessed tracks on the local device. Amazon has also applied these techniques for its set-top box, reducing video loading times by guessing what users will watch next.

Previous MEX sessions have looked at this issue from a couple of angles, such as:

1. Pathway #1 on network austerity, which explored how network-efficient design was a key element of delivering good user experience.
2. Pathway #14 on context, examining best practice principles for understanding user context and design experiences capable of responding.

Microsoft’s approach adheres to several of the best practice principles uncovered during the September 2012 MEX sessions on this topic, including:

1. If there is any doubt over user intent, do nothing without explicitly confirming. The experience design of these confirmations is crucial to overall success.
2. Contextual experiences should evolve the frequency and significance of their interventions gradually and not seek a single transformational moment. This builds user trust and reduces the risk of permanent failure and rejection.
3. Context awareness should follow the user and not become constrained to a specific device. Preferences and identity should be understood across all touchpoints in a user’s life, within the constraints of how they wish to manifest their behaviour and identity in different situations.

However, there were also several additional nuances to crafting good context aware user experience which those applying predictive approaches may wish to consider:

1. Users maintain an ongoing narrative around their lives. This is done at both the conscious and sub-conscious level, and elements of it are both projected to others and kept personal. Context aware services should sit within and support the shaping of this narrative. Understanding its intricacies should guide design.
2. Contextual response does not always require cloud connectivity. Much can be achieved within the confines of individual devices and sensors. Users may be reassured if their contextual experience occurs primarily at a local level.
3. Humans have innate pattern recognition ability. Designing to tap this trait helps users quickly confirm intentions when there is ambiguity over their context, with little additional cognitive load.
4. Context aware experiences should reflect the lies users tell themselves, not hard facts as understood by computers.
5. Users take the first use of contextual data as their benchmark. The way in which those contextual data are used should never be altered from that benchmark, at the risk of losing user trust permanently.

This ability to shape user experience in response to a constantly evolving picture of user context is becoming just as significant – perhaps more so – than the already well established practice of visually responsive design. It is also an example of how an individual MEX theme which was once explored as a specific topic grows into a set of underlying principles now considered fundamental to good user experience practice and embedded in the work of the MEX community.