Jason Sidabras

This is sort of a repeat blog, but I have a lot more reasoning behind it. Plus, I wanted a dedicated blog to this subject. Which subject? Well, if I had 60,000 employees at my disposal. This blog can also be titled: How MS can stay on top for the next 10 years.

Let’s face it, MS has gotten sloppy. Perhaps they always have been, but Vista’s “success” only shows that lack-luster OSes can be accepted on a mainstream level if the right people are involved. Let me give you two examples that bugged me on my trip to Colorado.

First: Vista is piss slow. For a laptop computer, which by default, that has a dual core Intel with a 1 GB of ram and (at least) a 128 MB graphics card I should be running circles around everything. Since Apple’s switch to Intel, people have talked about not seeing the infamous “beach ball”. (For those of you who don’t know what the “beach ball” is, basically its the PC version of the hour glass.) Why has this time indicator disappeared? Because stuff loads hella quick! But for MS, they have gotten rid of the hour glass for some swirling circle which appears on everything you do. Windows Button (no more “start”), searches and worse of all a 30 second pause before anything is opened. I assume the blame cannot be only MS’s. You can blame the AV companies for having to scan everything twice before opening also. If Vista is truly Windows 6.0 and was a complete rewrite, the mark has been missed. One can have tasty graphics without extreme overhead. For example Beryl.

Additionally, MS has added a little docking bar that runs from the top to bottom of your right side of the screen. This bar is translucent and defaults with an analog-looking clock on the top. I assume this can be turned off, but what is funny about it is it doesn’t allow clicking of icons behind it. So when, say, a presentation has been dragged behind it: double clicking results in the property box of the translucent window to activate. There is NO reason that this dock has to extend the entire length of the screen.

Second: Let’s face it… Powerpoint is a standard. Not a good one, but it is one. It has basically replaced transparencies as the #1 way to make a presentation. So why is it that Powerpoint and the ppt file is not standard compliance or its basic features not opened up?

Both Apple’s Keynote and OpenOffice have succeeded in creating a ppt read/writer, there is absolutely no guarantee that your presentation will look how you intended it on a different computer. This was the case in Colorado when the Vista machine was having trouble establishing a connection with the projector (not 100% Vista’s fault). An Apple computer with Keynote was put up to take its place. Guess what? Not a single equation existed. On a technical presentation about how Silicon Holes give EPR signal and how local B field deformities are measured, you can guess an equation or two was needed. Mid presentation the system was replaced with an audience members XP machine with better luck and alignment with the moons of Saturn.

Personally, whenever I get a chance I use LaTex and the powerdot class. It is missing some of the fancier features (animated gifs and screen wipes), but for reliability you can’t beat a PDF. (For windows users I suggest TeXnicCenter, I wish they had it for OSx, top notch.) It also has some features like a TOC on the left of your presentation which allows you to jump to different parts out of order and looks MUCH more professional than skipping through slides. Especially if you don’t want to bring up those slides AT ALL.

So on to some points

Microsoft Office

Let’s start with the easiest: Proprietary formats. We all agree they are bad. From DRM, aac and even mp3s to doc, ppt and xls. I understand there was a time where Word Perfect files could not be opened in MS Word for profit and that even MS Works could not edit doc files. This made sense in 1998. Where companies were trying to get the edge on the WYSIWYG market and no one knew which way was up. Now a days this is ridiculous. There should be no reason why MS has to get its own OpenXML Format, which isn’t open. It is an XML file with garbage in between. Instead they are “supporting” a third party translator between the MS XML to the Open Document Format, where ODF is an ISO standard.

So is this the death of MS Office products? It shouldn’t be. MS can do a few things to save their Office line: i) Create ONE Office product suite with full capabilities but sell “addons” for fancier outputs (also allow third party extensions not ActiveX or macro stuff), ii) support ODF fully and support legacy doc->odf transfer all the way back from Word95 (which only they can do) its time to kill all of the old files, iii) get the “basic” ppt and xls file format transfered to OD and leave room in the standard for proprietary enhancements.

Why buy MS Office then? Well this is where the billions of dollars of PR work will work and the “addons”: Idea: One is an IMAP like server on the new MS OS (detailed later) where your computer acts as a catch all for your documents. A personal .MAC account basically but not on MS servers. Idea: Streaming ppt from your computer at home with an Xorg-like terminal. This way everything looks exactly how you want it. This would be HUGE.

Microsoft Online Apps

Microsoft is going to try to stick their heads into this and they should have done it 3-5 years ago. Since Google started buying up companies and made itself the “king” of the internet, MS has been trying to catch up. Personally they can almost NOT enter this market. Especially for Office to be successful. But the two ideas above would take “online editing” to the next level. Google’s doc apps are great for collaborative writing but I don’t see them taking off exactly in that form. Too primitive. This is where ADAPT comes in:

ADAPT is the fictitious protocol I invented today. It stands for Any Document Access Protocol/Transfer. The protocol allows a kind of read/write which, as far as I know, has not been implemented in this way. The idea is a mix between svn, IMAP, ssh and Xorg tunneling. The ssh part comes in because the transfer HAS to be encrypted. I even suggest giving $10 million dollars to the OpenBSD team and use ssh (new MS OS detailed later). So obviously make the viewer free or part of the explorer package. This allows you to log into your computer from anywhere using a dynamic DNS server or your IP. If it isn’t part of the explorer package, I suggest making it cross-platform and U3 compatible. I highly suggest the latter.

Now you have a dedicated ftp/IMAP-like server for access. Every time you drag/drop a file into this special folder you must set its permissions. Users are added in an address book-like key server with a sha1 key associated with each person. Idea: Consequently this will cut down on spam and filtering because this key is used as a signature in Outlook/Outlook Express as a signing key, ensuring the person is who they are. Just like PGP signing, which will only come with the new Office “addon”. This then allows you to set a simple and advanced set of permissions ranging from simple read/write access to only allowing editing or highlighting/note taking. Now your co-workers and classmates can only see what they are allowed to see and other things just don’t show up. None of this “double click” to find out you can’t read it. You just plainly don’t see it. Saving the file saves on your computer with a full log of changes and svn functions for reverting back. Copies are merged and outdated just like svn and you can even use that stupid color changes that Office uses now to help merge.

Finally comes the Xorg tunneling. I am sick and tired of going to a different computer and seeing my presentation/office document/whatever differently because of versions or fonts being different. So lets tunnel the damn thing. Using the cross-platform/U3 compatible program I should be able to open up the file and use all of the libraries and fonts be it on a thin client at work or a laptop at a conference. We have the bandwidth and we have the technology. Fool proof and rock solid. To my bosses question of “What can you use in a presentation that you can make sure [fonts/equations missing] doesn’t happen?”, this would be the answer.

Microsoft Windows

Now to the crown jewel, the OS. This writeup will be similar to my previous writeup but with a bit more attention to detail.

Windows needs a rewrite that shows off what 60,000 of the smartest programmers can do. But this is not their fault. There are some horror stories in the MS blogs about features and buttons being added and taken away from the higher ups only to be added again by a different development team. The one that comes to mind is the Start Button Showdown. This hierarchy has to go first and a re-shuffle of the company must be done. Once you become like the oil and car companies you know you’ve made it, but now it is time to go beyond that “baby Huey” like company state. Companies shouldn’t reach critical mass and struggle to produce good efficient products. Instead it should be the opposite.

So here is what I suggest: One of two things: A compete re-implementation or a compete source sweep of the NT kernel. I only bring up the source sweep because of the work that has been done on the NT kernel up to date. My slight change of heart (more like pity) came from this article about what should be done for Windows 7. Where Thom suggests using the NT kernel because of its track record and years of abuse. But in the end he comes to the same conclusion I did (though he doesn’t explicitly say it): “Vista could be a success if they can prevent 0-day vulnerabilities” by creating patches quickly, often and before you have a reason.

But one prediction I have is that Vista isn’t “hacker-proof” it’s 0-day fix favorable. With the complete rewrite there are many additional tools that developers have come up with over the past 10 years (since NT complete rewrite) that makes managing code easier. If MS can keep up with 0-day vulnerabilities at a significantly faster rate than they can with XP then, Vista is a success.

True, Vista has only been out for 6 months but we have yet to get a patch “just because” or “could possibly” cause a problem. It seems it is always because problems have been found in the wild. (Unlike OpenBSD). To Microsoft’s credit Vista only has two unpatched flaws (18%). One is an ATI kernel driver DoS and one is a weakness in directory monitoring. Both rated low on Secunia and both local only. In the grand scheme of things this is promising.

One thing MS won’t do with their NT kernel is open source it, and that is why I suggested a FreeBSD fork, named Dembski (a joke no one originally got or thought was funny :p ). Taking a lesson from Apple, the GUI would not be part of the OSS Dembski, only the core OS and most drivers. It should be MS’s duty to make those drivers blob free and use its clout to push for a more open source driver world. In the end this will make MS far more respectable in the OSS world and probably draw many developers to Dembski. I suggest Licensing to be completely open, but some restrictions can be applied. Like “non-commercial use” only. Although I don’t advocate that. As I mentioned before, stating reasons like Apple’s wireless bugs and among others would be good PR for the removal of binary blobs.

The new Aero interface, which is hopefully re-written using Xorg libraries and proprietary libraries, is also not supported in Dembski. This makes sure people will buy the new windows for that “New Windows Feel and Look”. The Xen support would be built into the kernel but the official WindowsXP image supplied with the non-OSS version will not be included. This will lead to pirated WindowsXP images, but at this stage in the game it shouldn’t matter. Allow this image to be refreshed anytime the user wants to get rid of potential registry problems and put the thing in a jail. It seems OpenVZ is the way to go with the features and throttling available for the system-level virtualization.

In Dembski, allow for a “talkback” to the MS bugfix servers. You then have a constant stream of systems to help facilitate trends and problems that may rise. NO PERSONAL INFORMATION, but plenty of computer information. If the port becomes blocked by a firewall, kindly (annoyingly) remind the user to unblock the port as part of their user agreement to use Dembski. Send system information once a month and after each patch. Patches should be regular and tested on Dembski first, unless a 0-day vulnerability is noticed. It’s best not to abuse these users so make sure it is stable before sending it out. User groups and the OSS community will provide support and all Dembski computers will have “beta codes” to make sure MS support people’s time isn’t wasted.

Finally META data is a must. It has to be painless and fool proof. Spotlight is getting there but it needs some more work. This may be in the works with WinFS but what is needed, and as far as I know, is missing in OSx is portability. the .DS_Store directory gives a significant amount of details about the files stored in the directory. This seems like the wrong way to do it. Personally each file should have the data stored in its headers, similar to an mp3. This way the portability is there.

Finally

The most important thing MS has to remember in the next 10 years it it will have to be nimble and be able to change with the times. The above mentioned ideas are ones that I think are important in the coming years and will set up MS to stay on top. MS is the only one with the initial infrastructure to pull it off without hurting the bottom line. The applications google has been coming out with seem like hacks on-top of the MS OS. To be able to trump google the only way is to make your product better and less prone to people needing the “oooh cool” or “why didn’t this do this before” upgrades.The virtualization of WindowsXP will allow for people to have to upgrade to a new chipset and MS to have temporary backwards compatibility until the next OS installment.

Other flagship products like Microsoft Office Suite need a serious overhaul. The newest versions lend nothing to innovation and in turn will lower sales. By creating the ADAPT protocol you can create something that no one has done before and would make the “online apps” obsolete before they even took off. Of course there will be niche markets where online applications will work, but in their current form the office products don’t seem like one of them. Collaboration can be done at the file system level and without dealing with multiple copies everywhere. Gates wants everyone’s computer to broadcast itself and be accessible. I agree, but you have to be sure that the system you put out for everyone to see has to be secure by default. Otherwise your a sitting duck and there are millions of strains of avian flu waiting for you.

Leave some comments, I’d like to know what you think about the ideas. Especially the ADAPT protocol. If anyone is interested in creating this protocol let me know. I would like to participate or at least get my name on it.

You can follow any responses to this entry through the RSS 2.0 feed. You can leave a response, or trackback from your own site.