Perhaps the buttons don’t look like buttons? Wrong colour? Maybe they need a hover colour-change? Or change the “Nope…” text?
Was wondering whether in future we include a machine translation tool - but some of the text is so “informal” that it might be challenging anyways. (For example in our first data dive, I learned what “impazzito” means in Italian.)
Indeed yes, in fact there is an interesting result regarding a Dutch text problem in the FaultCat results already.
Google Sheets have recently implemented language detection and translation functions which we could use to enhance the devices table in the Fixometer.
So here is the first set of results. The column headings are explained below. The query is brand new, its all one big voyage of discovery! 
Click here for full view
- Event Opinion : the opinion that was recorded at a Data Dive event for this record (Note that “Unknown” can also mean the record has been entered since the last event which was early September 2019)
- Event Opinion Majority: Does the majority of crowd opinions agree with the event opinion for this record?
- Crowd Consensus: Do all of the crowd opinions agree with each other? “-” indicates only 1 crowd opinion logged so far for this record
- All Crowd Opinions Count: How many crowd opinions have been logged for this record
- Top Crowd Opinion: Crowd opinion with the most “votes” for this record
- Top Crowd Opinion Percentage: Crowd opinion with the most “votes” as a percentage of all crowd opinions for this record
- Unique Crowd Opinions Count: Number of unique opinions logged for this record
- Unique Crowd Opinions: List of of unique opinions logged for this record
- Category: Product category that was recorded at the event
- Problem: Problem that was recorded at the event
A few interesting results pop out at me:
- Device ID: 7393
- Event Opinion: Integrated keyboard
- Event Opinion Majority: No
- Crowd Consensus: No
- Crowd Opinions: 3
- Unique Crowd Opinions: Integrated keyboard,Integrated screen
- Category: Tablet
- Problem: Defective home button - Advice given on how to source replacement home screen (4 years old)
The word “button” in the problem text above suggests a physical object and so a keyboard fault. However the category “Tablet” suggests perhaps a virtual button therefore screen fault?
Another interesting one… lost in translation?
- Device ID: 13428
- Event Opinion: Ports/slots/connectors
- Crowd Opinions: 2
- Unique Crowd Opinions: Operating system
- Problem: connectie hoofdtelefoon kapot (translated “headphone connection broken”)
So what do we do when the crowd reaches consensus but the fault_type is very clearly wrong?
What a difference a word makes here…
- Device ID: 1273
- Event Opinion: Configuration
- Event Opinion Majority: Yes
- Crowd Consensus: No
- Crowd Opinions: 3
- Unique Crowd Opinions: Configuration,Integrated optical drive
- Category: Laptop medium
- Problem: Laptop - DVD driver corrupted
One person saw “DVD” but not “driver” which makes the difference between “Configuration” and “Integrated optical drive” but would also make the difference between “Software” and “Hardware” (not that we’re collecting that with FaultCat).
My takeway from just those examples are that FaultCat could prepend the Problem text with the Category instead of it being at the top of the page, maybe even the brand/model if they are present.
Also to take a look at language detection and translation - even if its just a link to Google translate to begin with.
I do think including the brand and model data would be helpful, it’s not always available or accurate but would give some useful context. I clicked through 10-15 examples on FaultCat and at least one made me wonder whether the category was wrong (“Laptop small” but description was “Personal DVD player - not reading discs” - should this be “Handheld entertainment device” instead?)
3 Likes
We’re doing some exploratory work with product categorisation at the moment and toying with a microtask app to sanity-check the existing categories and perhaps explore different methods of categorisation.
1 Like
I have put a new version up on a test server to see what we think. It includes some of the UI and functional tweaks discussed above.
- Added translate button (just opens Google translate in new window but fills in the text)
- Moved category and repair status into the
problembox and added brand and model - Added a “Nope” button that unhides all the category buttons - perhaps restricting the initial choices might make the usage more obvious?
This version is not collecting live data so feel free to wildly click random buttons. Any features deemed to be an improvement can be integrated into the live version.
1 Like
Just tried another 20 or so. There are a few where the description is a good indicator but there is no obvious category. For example ‘3G not working’ on a tablet, could be the internal aerial. There is no obvious category for things like internal aerials. These can often get disconnected or damaged on previous repair or opening up.
“Internal damage”? This fault_type arose from things such as devices being dropped, liquid spillages and “user anger” (as recorded in a couple I’ve seen).
Which reminds me that there isn’t any way of distinguishing faults that are caused by poor product design/manufacture and user misuse/abuse/accident. Should the blame for breakage always lie with the manufacturer? Debatable in that devices shouldn’t be “easy” to break and its not possible to make devices 100% waterproof (and still affordable) but if somebody jumps up and down on their laptop and we record it with a fault_type that describes only the damage and the assumption is poor design or manufacture, then, hmm…
Looking good and like the translate button! My only question/critique is that “Nope, let me pick something else” should probably be another colour, like the charcoal colour it reveals when you click? Otherwise the two red buttons are slightly confusing.
Also wondering, given @Fabrice_Arfi’s confusion initially whether the help text “A [device] with repair status [x] had this problem: [x]” is still needed? What do others think?
1 Like
In the spirit of festive democracy I’ve put up a microtask to get opinions on combos of button colours.
Random combination of Bulma button colours presented on each Fetch. Select Yes to log as many combos that you like. Selecting No does nothing, its just there to complete the template.
Might also mention that underlined letters denote keypress options, e.g. Press keyboard ‘F’ for Fetch.
Completed a number of task button tests.
How about a simple traffic light system?
Yes / probably
I don’t know
Nope, pick another…
4 Likes
Can try traffic lights, its a nice simple concept. 
2 Likes
