Ah thatâs a good explanation. In that case itâs working very well particularly when thereâs just one match it saves a lot of work finding the category.
I know this is no use for touch, but where a mouse is being used could you provide a tooltip/hover on each selection? In particular there is a distinction between âunknownâ=>âI havenât the faintest idea what the problem isâ and âotherâ=>âthereâs a specific fault which doesnât have a categoryâ which could be explained with hover.
âLiquid damageâ seems like it would be a useful category.
My favourite problem so far is âDod had bitten on itâ - unfortunately thereâs isnât a selection for âBitten by Dodâ (hopefully a dog not the ownerâs Dad) so I was going to choose Unknown but the most obvious thing to have broken is the screen so I chose Screen. Now I see that fault is one of the ones with conflicting opinions Screen/Unknown/Other. Could you change âOtherâ to âKnown cause but no categoryâ or âCause with no categoryâ to make clearer that it doesnât mean âthe fault cause is unknownâ (which is what Unknown means)?
Tooltips are a great idea! The definitions could go into the âinfoâ modal too I guess.
FaultCat had âinternal damageâ to cover things like dropping devices and accidents with liquids (and âuser angerâ). The options for mobiles are designed to map to a set of commercial repairersâ fault types and they donât have anything with a generic âdamageâ type. Youâll notice that with the exception of âUnknownâ and âOtherâ each fault type is either hardware or software related (except perhaps âstuck bootingâ which could also be firmware). I think the aim is to target specific components.
Also, prior to confirming the given list of fault_types I ran a term frequency query across the mobile problem text and found only single occurrences of the words âliquidâ and âwaterâ so a specific fault type related to liquid was ruled out in order to keep the list concise. Of course other words like âoceanâ have turned up! (Possibly toilet, sink, coffee, coke, wine etc.)
With regards to more granular fault categorisation, after MobiFix has collected enough majority opinions we could re-run the âOtherâ and âUnknownâ ones with more specific options - if there are enough to warrant doing so. Or we could simply share a spreadsheet and let interested parties enter opinions there.
There are a few other items in the list we may look at refining such as splitting screen problems into âbroken glassâ and âtouchscreenâ if possible. Apparently there is interest in knowing which particular camera was at fault also.
When all the votes are in weâll see what else could usefully be achieved.
Yes I see what you mean. Might try some sort of tooltip as suggested by @Ian_Barnard, there is always a bit of a trade-off between precise labels and keeping the screen real-estate tight and responsive.
Tooltips are live now! Thanks to @Ian_Barnard and everyone who supplied feedback.
Not every button has a tooltip, quite a few of the labels are pretty self-explanatory, but hopefully the new tooltips will help unravel a bit of confusion for some.
Thatâs strange, its working OK for me. Are you signed in or not? Tried a force-refresh of the page? Are you getting them just for MobiFix or have you tried other pages?
The only thing that changed was adddition of tooltips afaik. Iâve asked @neil to take a look in the logfiles.
EDIT: yep, heâs spotted something, only happening on certain MobiFix records. Iâm having a look now.
I actually have a question. A few of my friends said they are hoarding their broken old phones which wonât turn on. They donât want to throw them away because they have valuable metals and rare earths etc. But they donât want to put them into electrical recycling because they worry about where it ends up and about their personal data (they canât wipe/reset factory settings because the phone wonât turn on). Now, they also donât want to pay a lot to a repair shop to wipe it. Whatâs the best way to go about this wiping? Will your SIM provider do it for free?
One minor buglet asking for translation of sembla que allďż˝ on sâendolla estďż˝ xafat i no puc carregar la bateria gave a blank page on oogle translate.
One suggestion for future things like this - might be useful to put a record id on the page so itâs possible to refer to e.g. ânumber 1201 has a problem with translationâ. Also maybe mobifix is testing my consistency but I seem to return to the same record relatively often and having an id visible would either reassure me thatâs not the case or confirm my suspicion IYSWIM.
Ah, that string has some non-printing characters which would explain why it was not an acceptable param in that URL.
Iâve been thinking that too.
Indeed these early microtasks have no way of checking for duplicate presentations but it is on the card for work this sprint and the next. Although tbf there are a lot of mobile records that appear duplicate but actually arenât, they just have the same brand/model and problem text. I ran a few queries and found that I gave 113 opinions and saw 2 of the same record 2 times. The more opinions you give the more likely you are to see dupes and as we get fewer to choose from - records with majority opinions are eliminated - you are also more likely to see dupes. Working on testing an idea to workaround that.
Most people give way less than 100 opinions, about 4 people have given in the region of 200 and one person has dedicated their time to about 850!
Just saw âneeded new connector - professional helpâ twice in a row, but maybe there are two of those? Also had â0.2kgâ twice today (answer: Unknown) and IIRC several times before. I guess the priority should be to ask me about the record with fewest opinions (or randomly if there are several with the same lowest number of opinions), excluding those Iâve already given an opinion on (as thereâs little point asking me twice about a record). That means you have to record users who have given an opinion on each record, or maybe when it hits 3 then you can erase the user IDs?
There are a bunch of these with the weight in as problem text, they come from the same group years ago.
This is how all of the microtasks currently work - âfetch a random record that has not yet gained a majority of opinionsâ.
This is not something the microtasks cope with currently, see below.
The majority of microtask users appear to be anonymous - either unregistered or not logged in and analytics shows that its not uncommon for users to start it anonymous and then log in. So Iâm going to try collecting the record idâs seen by each user in their session variable and then exclude those idâs in the fetch query. Ofc this means that you might see the same record if you change device, IP address, logged-in status or if the session expires (I think the session duration is quite long though for the same device/IP). However, it will be the next microtask not this one (yet). If successful it could be retro-fitted to the existing apps.
Should also point out that the user has the option to not give an opinion and âFetch anotherâ if theyâre sure itâs a duplicate although Iâm usually too in the zone to remember to do that myself!
A majority is only 2 matching opinions out of a maximum of 3 opinions. So any record that has yet to achieve 2 matching opinions can be shown to a user. For the record, MiscCat had a max 3 opinions and FaultCat went with max 5. We are as always assessing our choices with regard to the parameters.
No worries, we all certainly appreciate all of your efforts and feedback! I love the discussion and the arguments here and have certainly taken a lot of it on board!
I believe there are also credible corporates that wipe phones, but it requires some investigation. It depends on your personal risk threshold. Do you want data deleted immediately? (For example, weâre not certain that Mazuna wipes mobiles in the UK or merely does a triage here and exports them for data deletionâŚ)
Restoring to factory settings will hide these from an opportunist but much or all of the information may still be recoverable with forensic tools, which criminals may well be in possession of.
We may not want to say this, as not sure this erm, effectively allows for reuse and recycling. Also, erm, lithium battery? @philip
Restoring to factory settings should then be effective, but even so, if youâre worried you might want to finish the job with a hammer!
So, we are done! Big thanks to all who particpated!
Weâve not achieved a completed microtask before so I guess it was inevitable that the final state is not handled and now loading the page results in an error, oops! Lesson learned!
Weâve got a little work to do to adjudicate , analyse and report