DivX Development Direction - Re-Focusing on User Needs

Summary of Post:
This Management Consultant's report details:
1. Issues re re-focusing on simple bugs and user interface, rather than continual expansion into newer/better underlying code.
2. When Less is NOT More - how Converter's interface needs to be expanded a bit.
3. Get The Basics Right - fix the few bugs in Converter that have existed for years.
4. Make It Robust - Show status in ALL stages of Converter to avoid appearance of app hanging.
5. Make It Easy - Store customer 'last used settings' and/or customer default settings.
6. Don't Force Your Frills on Customers. The whole menu thing was way too buggy to force on everyone, and besides a large slab of customers didn't want it.
7. Set the CPU priority of all conversion/encoding to 'Below Normal' to give users back their PCs.
8. Return to basics with processing of all video in a given sub-directory.
9. Add audio level normalisation, just like the good MP3 software does.
10. Make the DivX website suitable for different classes of users.
11. Focus on what is important... retaining customers over exploring new frontiers.
___________________
I have thirty years experience in software design. I was on the team of five at Harvard Business School in 1978-9 which developed the first electronic spreadsheet ('VisiCalc' - the acknowledged pre-cursor to Lotus 123 & M$ Excel) - though I was only on design review, and my classmate Dan Bricklin deserves all the credit for the invention of the speadsheet. But I was also a Harvard Consultant to The White House on IT, and co-founded a large health software business.

As a user of DivX for a few years, let me make a few consumer-oriented observations about the direction/focus of software development at DivX, with some special focus on Converter. It constitutes a full list of urgently-required work to significantly improve the end-user interface:

1. EVERYONE IS TIME POOR.
Yes, the natural response to this feedback will be "we can't do everything - we have limited resources". But that is the crux, in a situation where resources are limited, what should be worked on? I suggest that DivX's development efforts appear (to an outsider) to be driven ENTIRELY by the desire of engineers to excel in some particular pet field of their expertise, and SHOWS LITTLE COMPASSION for the frustrations encountered by the typical user. I am an engineer, and know too well how this happens. M$ is an example - many users would exclaim "That problem would have been fixed long ago if they'd only asked customers what they wanted!" An example is that MSDOS from 1984 flushed out the directory updates to a floppy immediately after you wrote information onto a floppy, so you could remove the floppy as soon as the drive light went out... whereas 23 years later, WinXP does not flush out its directory information to USB-connected hard drives, even if the CPU is completely idle... you have to pre-advise WinXP that you would like to remove a removable media! Well, in precisely the same way, DivX has failed on a plethora of little ways to assist the 'typical user', instead pandering to the needs of the most advanced user, or latest niche to be addressed. But if you want to keep broadening the base, you need to make it better/friendlier, not focus on more speed/resolution. Otherwise a competitor could come along with simply a more robust/easier-to-use product, and your market lead will evaporate, expecially if it is a big player. You may think you are resource-poor to make the experience better, but your millions of customers are even more time-poor to learn work-arounds for the types of problems you could have addressed with friendlier application software.

2. WHEN LESS IS NOT MORE.
Before Converter was released, DrDivX did a reasonable job. True, it didn't test for conflicts etc, and was reasonably buggy, but it worked, and it had a usable/standard GUI. Then some senior engineering zealot within DivX pushed the line "we just want one button" and Converter was released. Well, the uproar of protests from existing users was so painful that DivX HAD to bring the Doctor out of the forced early retirement. Even now, Converter should be given a minimal pull-down menu just to comply with Windows software standards. There should be a 'Help|About' to at least disclose what version number you are on, and hell, if you are going to let people in on that info, you may as well allow them to 'Check for Updates' as well, as people may have been busy or on too slow a bandwidth connection when the last 'Update?' question was asked of them... and there is no way to revisit this if you ignore the opportunity.

3. GET THE BASICS RIGHT.
One of the most glaring issues regarding Converter is 'If you reduce an application to just Drag-n-drop and then one button, those two functions should work properly' but in fact they don't. Two years on and the Drag-n-drop still doesn't handle multiple files being dragged... and it doesn't even report the problem, it just mis-behaves. If you put a competent software engineer onto that problem, he'd have to have conquered it within a day!

And if you click the 'Convert' button, you get cryptic feedback about percentage done, which is rubbish. Even I have aborted large jobs when I first saw Converter sitting at “50%” or "100%" when it then didn't move off that figure for an hour or two... When you have jobs that run for hours, you need good status feedback that it is still working, and Converter does the worst of both worlds: It appears to show a status that changes, but this display is only updated during some stages of the process... which is worse than if it gave no status whatsoever.

4. MAKE IT ROBUST.
Clearly there are a number of codecs/filters/components which DivX needs to have present, and others which if present will prevent the proper operation of DivX. When you want to make software 'idiot-proof' for a wider market (ie beyond geeks), you need to do a start-up routine which first checks for the existence of these pre-requisite things (and non-existence of the proscribed things). Any issues should be reported to the user, with maybe a 'Don't show this message again' tick-box option. It is ludicrous that DivX promotes a third-party-provided 'Fix-all' piece of software which seeks to separately remedy the types of problems encountered! Most users will give up on your software long before they find (in FAQs) a link to another piece of software designed to check why the first piece wouldn't run.

Then have EVERY step in the process (eg Converter) report to the standard display routine as to progress, so it never appears to hang 'half way through'. This problem derives from not forcing all developers to use standardised routines for reporting/display. If your testers used real 3GB source MPEG2 files producing 600MB resultant DivX files, it would stand out immediately when there is too little status reporting. I can almost bet that the developers use tiny test files, and their tests fly through the process so quickly, that the lack of appropriate status reporting never becomes obvious to them.

5. MAKE IT EASY.
DivX has confused software that has a simple look with software that is simple to use. Nothing could LOOK simpler than Converter, but there are heaps of really simple ways to make it far easier to use. Go back to the old way of getting a few different types of users to log just how many keystrokes and mouse-strokes it takes them to achieve a certain typical usage of the software. Then look for what is highly repetitive, as that is the type of task a computer was intended to address. One thing that has always stood out to me is the arrogance of Converter remembering the DivX team's preferred settings, but refusing to remember MY preferred settings! Converter has certain standard settings, which is fine - but what I do is use 'Home Theatre' as I start with full PAL resolution (720x576) and don't want the resolution reduced, but then I pull the resultant video bandwidth back from its Home Theatre default of 1300bps to 800bps, as that is actually fine for most TV viewing... but do you think that after two years I can get Converter to remember my preferred setting? No! The good doctor used to at least remember your last setting, but in Converter, you have to display the list, and then manually adjust each item in the list to your preferred default settings... which is a huge waste of time and very frustrating. Hell, at this point I'd be happy for Converter to remember my setting even within the one session (ie in RAM) but clearly it would be neater to save it to disk for use as a default in my next session as well. Again, I'd bet that a competent developer could save a customer default/custom setting(s) in less than half a day. And even if you didn't want to change the software at all, you could at least add an extra default 'Small Filesize Home Theatre' vis-a-vis 'Larger Filesize Home Theatre'. But I say, 'Count the keystrokes', and you will learn what ridiculous time-wasting you are causing your customers. Defer the next beta of an improved engine for a week, and fix the car body around the engine!

6. DON'T FORCE YOUR FRILLS ON CUSTOMERS.
OK, so someone in DivX Labs has a passion for bringing out a competing technology for the types of menus you get with DVDs. I'm really pleased for him/her, but PLEASE don't force your 'frills' on me. I've found the quickest video editor is the one that simply cuts from nearest reference frame to nearest reference frame, without any re-rendering. This is typical of the free software supplied with many low-cost Video-to-USB capture devices. But if you take a free-to-air TV program and cut out the ads, the resultant MPEG2 file has serious audio sync problems, proportional to how many ad breaks had to be removed. But the individual segments/cuts are fine. I used to be able to do the concatenation (joining) in Converter, saving having the video editing software do the concatenation. I'd simply drag each video segment (as Converter can't keep the order straight if all were dragged conjointly) to Converter and tick the 'Merge into single DivX' button on Converter.

Now, I am simply not allowed to do that simple concatenation in Converter. Since Converter introduced a new feature of being able to offer a 'menu', I am forced to use that menu system, even if I want my viewers to simply see the footage seriatim (in sequence). This approach runs contrary to all customer-centric marketing logic. You can ask “Do you want fries with that?” but if you force fries on all your customers, you will certainly upset a sizeable segment that never wanted that side-dish! The proponent of the menu system was either too much of a zealot for their own work, or too lazy to update the user-interface to include an option for the new feature. What should exist is the tick-box to 'Combine videos into a single Divx file' and then under that, normally-greyed and indented a little to the right (ie conveying it is an option that only makes sense if you HAVE elected to merge clips) should be a second tick-box 'Add menu'. It was simply wrong to join these two questions into one. It failed the 'reverse compatibility' test and the 'customer is king' rule.

And for those who think I am making too much of an issue about not being able to 'opt out' of the latest frills, let me point out that the menu system foisted upon all users was a lot like a v1.0 of a new M$ OS – it just didn't work. First it made the output file sizes twice as large as they had been, due to a silly bug. Then it meant that your viewers could not simply click on your video to play – they HAVE to look at the menu and made decisions. If you put a number of DivX files on a DVD-R and play them in an approved DivX-playing DVD player, you can no longer have them flow into each other (as used to work) as each time you now encounter a new file, the user has to do some user-interfacing, to elect to start at the beginning of the file (which I'd have thought could be a default). Anyway, then in an 'upgrade' DivX decided to increase the point-size of the menu text, causing on-screen truncation of the text, which in my case put the Ch1, Ch2 etc off-screen, leaving viewers really confused as each menu item looked identical to all the others. As usual there was no disclosure as to the limit in terms of number of characters for a menu item. Now, in a further 'update', the compulsory menu system is now fully broken. With the latest update, you get a resultant DivX where if you chose 'Start' (ie Chapter 1), it NO LONGER runs Chapter 2 etc.... you have to play each scene/chapter separately, which is so hideous in terms of changed user-interface that it is worth avoiding at all costs. And if the bugs weren't enough to make you pray for the previous working version, there was the incredibly inefficient software coding. The menu system foisted upon you now required another whole pass through the temp output file before final writing, slowing down the whole process to provide the feature you didn't want. And with the 'stage and percent done' display not updated, it simply looks like the software has hung at this penultimate step.

The lesson is clear, make any new frills tick-box options, so people can ELECT to experiment with your new functionality while it is buggy, but don't force your agenda or perceived customer needs onto all customers. All I want is the ability to concatenate again, without any damn menu.... but if you are going to offer a menu, you probably need to let users decide the point-size, or make the software smart enough to determine the point-size to fit the provided text within the screen area!

7. MAKE CONVERSION A BACKGROUND TASK.
Anyone doing MPEG2->MPEG4 conversions would be happy to still have their PC available while the work progresses. Converter et al should DEFAULT to Windows 'Below Normal' task priority. I routinely go into task manager and set the priority down after I start a job, but that is another waste of time... Good programs like Diskeeper automatically do all of their non-user interface work at below normal priority, leaving the PC available for other uses. Many average users don't know how to change task priority, or wouldn't bother... but it would be simple to do and give everyone a better experience. And if you are in the 1% who are in a rush for a conversion, then you wouldn't be doing other processor-intensive tasks on that PC while the conversion job is active anyway (so it makes little difference to your completion time if Converter defaulted to a lower initial CPU priority).

8. JUST PROCESS ANYTHING IN A DIRECTORY.
But of course, once you make it a low-priority task, it may as well sit in memory (like Diskeeper). Then if you supported a user-settable default encoding, a user would simply have to direct the output from their preferred video editor to put the MPEG2 file into a known directory, and then the Converter (by checking every few minutes) would see there was a file to be encoded and would simply jump to work. That way, if you have another video edit finish during that encoding run, it would simply be picked up in turn and encoded. Currently, you have to wait for Converter to finish its last batch and wait till you are awake the next morning to give it its next chore. This way they would be queued in the simplest way possible - a sub-directory. And you can't mess up a drag-n-drop into a sub-directory, whereas DivX still can't quite get the code right to enable a drag-n-drop of a list of files into its application... so this approach would also fix some long-standing bugs that have never been addressed. Of course directory-processing is getting back to the type of capability DrDivX had... And when you pick program-default directories, NEVER pick a directory under 'Program Files' for large video files (ie user data) as DrDivX did, as you really want your data files well separated from the Operating System (given all the problems one can encounter with WinXP). Besides, you want to be able to back-up your data separately from your OS. And you would need to allow specification of external drives, as all video processing works so much faster from/to external drives, because your 'C: drive' temp files are on a separate physical drive from your primary input and output files!

9. ADD AUDIO NORMALISATION.
As audio level problems persist, rather than refering users to run some other filter software, why not simply put a tick-box into Converter/DrDivX which, if ticked, will do audio normalisation, just the same as the better MP3 music handling software will do. In other words, rather than trying to have thousands of users set manual audio level controls, just have the software do a few samples throughout and then adjust accordingly to give a 'normal range' audio amplification. That way one COULD eventually listen to DivX videos on phones and other mobile devices, as the platform would not be ruined by all manner of encoding levels of audio. It is a very simple idea, but would work well in practice, as no-one wants clever video encoding, but audio they can't hear, and who wants to manually set the volume for everything. It should be like TV (without the ads) where you don't have to keep adjusting the volume level as the program changes from one to another.

10. THINK OF DIFFERENT CLASSES OF USERS AT DIVX WEBSITE.
The DivX website caters for first-time sign-ups and for media downloads. It does NOT cater for existing customers re-visiting for software updates. For one thing, it does not state on or near any of the 'Download Now' buttons what will happen if you already have a current licence for that product. Previously there was a small link “If you already have a licenced copy, click here” but that has been removed. You need an explicit description of whether clicking on any particular download will wipe any existing licence, otherwise users will be justifiably worried about clicking on any 'trial' version. YOU may know what will happen, but your userbase are not appropriately informed.

Also, the DivX Labs website is a great decision. The DivX website had no way of providing technical feedback to the company. My own take on this was that, if you give people no way to provide feedback, that fact itself is prima facie evidence that there are too many bugs in your software, or it is too hard to use. A viewer was only ever directed to the FAQs. Even if you don't provide technical support, you should have a web-based bug report submission form, noting that you may not get back to users about it, but thank them for making the effort. I was the one who discovered the doubling of output file sizes as soon as the 'forced menu' supposed-upgrade was released. I looked at every link on the main website but could not find how to report a bug. Eventually I had to look up 'Investor Relations' and emailed someone in Finance who passed it on - and he emailed me back, acknowledging that mine was the first report of that problem, and that it had been fixed promptly. It shouldn't be that difficult. It reeks of a company not wanting to hear of problems.... which is often a portent of a company which has lots of unresolved problems.

11. DON'T WAGE WAR ON TOO MANY FRONTS.
As part of the issue of not having so many problems unresolved, just don't wage war on too many fronts simultaneously. Anyone who has played the board game 'Risk' knows that if you have your troops spread out over too many fronts, you will not win world domination! Similarly, it would be better to do the mobile device market a little slower, and ensure that your mainstream PC and DVD-player products are robust and do what they claim to do, with easy user-interfaces. None of the things I've suggested would take more than a day's development work, yet all would make use of the DivX software far easier.

I like the DivX company; generally like the DivX standard; can just tolerate the poor user interface... Indeed I am sufficiently supportive to take the time to set-out what I think should be development catch-ups... But if another company came along with a good MPEG4 suite that was friendlier and more robust, I and many others would jump ship at this point, as DivX hasn't yet quite migrated to a professional quality-assured software house! AND it is missing a critical feedback loop. It needs someone really powerful within the organisation to represent (be a vocal advocate for) the userbase as a driver for gradual improvement in the existing products, to avoid having the development team put ALL their efforts into the latest new diversion. Great companies ensure their main product lines remain solid and well in tune to customer needs, even while the cultivate new markets.

I am happy to receive any feedback on the above issues on my email address of “prof at-symbol post.harvard.edu”.