Why Microsoft is right company to make a next gen flight sim

[abrams_tank 7,425]

12 minutes ago, Bernard Ducret said:

Yes, MS sells this product under its brand and financed research and pre-release expenses, and continues to do so today, but my old school of thoughts reminds me to give credit where credit is due.

Asobo deserve credit. As do Working Title. But my post was more about who can provide the resources to make it possible. 

But yes, I agree with you that Jorg did a very good job in scouting Asobo.

[abrams_tank 7,425]

1 hour ago, UrgentSiesta said:

Google's also highly focused on consumer entertainment like Apple, ad revenue, etc. And their biz services divisions are doing rather well, too. But they can't seem to find their a** with both hands when it comes to gaming, though they have all the other pieces/parts for flight sim, unlike Apple.

I agree Google has all the key components to make a very successful next gen flight simulator if they decided to emter the gaming marker. 

The other wildcard is Amazon. Amazon has many of the components too like the AI tech and AWS servers. And Amzon is also into gaming, although I read they recently laid off people from their gaming divison. I'm just not sure how good Amazon's satellite & photogrammetry data is, even though I see they offer satellite data services.

Edited February 23 by abrams_tank

[David Mills 4,773]

Jorg said they'd be releasing more info about MSFS2024 in Spring. Spring is now less than four weeks away!

[UrgentSiesta 931]

3 hours ago, David Mills said:

The current advancement of flight simulators reminds me of the advancement of physics in the early twentieth century. Back then, a solitary individual (e.g., Einstein) -- or a small group working together (e.g., the Copenhagen School) -- could make revolutionary discoveries through pure genius and a pencil and paper. Billions of dollars were not required. Likewise, early flight simulators were the work of a relatively small, brilliant group of coders working on limited resources (e.g., Sublogic, LR). Considering these limited resources and technology of the time, their flight simulators were impressive.

Today, however, new discoveries in physics require equipment costing tens of billions of dollars (e.g., the CERN particle accelerator, the Webb Space Telescope). Likewise with flight simulation: Further advancement in flight simulation can only be achieved by companies with ten of billions of background assets for support and development (i.e., Microsoft).

Good thoughts.

I think you're right in re AAA title-ish sims/games that want to appeal to an extremely broad swath of potential simmers/players.

But, as with Laminar Research, 1CGS (IL2), Eagle Dynamics (DCSW), AeroFly, Boundless Dynamics (VTOL VR) and even BMS (torch bearers for the modernization of Falcon 4.0), we see that small(er) shops can still make first rate sims - when they aren't trying to be all things to all people. eta: funny thing that Prepar3D didn't occur to me, but it's another example of a sim that's (slowly) getting better over time. Kind of an edge case due to their core audience, tho...

And we see that they can keep substantially improving their already genuinely good sims, and (thanks to market pressure from MSFS) at a rapid rate, too.

And let us not forget that though MS/Asobo have/are delivering unprecedented improvements in general purpose flight sims, there are areas / features that they still need to improve upon, or outright add to the sim.

I guess in terms of "billons and billions" (cue Sagan 😉 ) being required, I think a lot of that has to do with their chosen value proposition and modus operandi for the sim itself.

Edited February 23 by UrgentSiesta

[UrgentSiesta 931]

3 hours ago, Bernard Ducret said:

Where is Asobo mentioned  in your discussion and where is there any credit(s) for those who created the sim that we are using? Granted, MS is the financial and marketing strength that made this sim achievable, along with the Bing imagery (but we know of other equally good if not better data sources...).

MS hired Asobo, most likely, after intensively searching the market for a talented group of developers, and Asobo were chosen for a reason (actually many). So why not integrate into your brilliant demonstration some words of thanks to the people who for the past four years - and probably as many before the release -  developed and improved this superb technological undertaking that we all enjoy?

Yes, MS sells this product under its brand and financed research and pre-release expenses, and continues to do so today, but my old school of thoughts reminds me to give credit where credit is due.

IIRC, I think it was somewhat the reverse: Asobo built a "digital twin" world sim, and they were using a TBM to show it off, and then somebody took it to MS and the (now obvious) conclusion was reached...

[Sky_Pilot071 580]

40+ years experience.

sp

[Ray Proudfoot 8,961]

3 hours ago, David Mills said:

Jorg said they'd be releasing more info about MSFS2024 in Spring. Spring is now less than four weeks away!

Meteorological Spring is 7 days away. Astronomical Spring falls on 20 March. Ask him which spring he’s referring to. 😉

[David Mills 4,773]

2 minutes ago, Ray Proudfoot said:

Meteorological Spring is 7 days away. Astronomical Spring falls on 20 March. Ask him which spring he’s referring to. 😉

Hopefully he was talking about Meteorological Spring. I'll definitely ask him next time he stops by the house -- probably later this evening. 🙂

 

[eslader 1,784]

14 hours ago, abrams_tank said:

Unless your company owns this technology and has the full rights to use stuff like this, it can get expensive

Your premise assumes the actual AI processing is cheap enough for a huge company like Microsoft to be able to offer it for free. It's not. ChatGPT spends a fortune to run itself. Microsoft would have to do the same. The supercomputer MS built for OpenAI has 285,000 CPUs and 10,000 GPUs. 

The electricity costs alone for running a system capable of supporting a LLM AI are ruinous. If Microsoft did their own version of BATC natively in MSFS, they'd have to charge us too.

That will probably change in the future. EINIAC consumed 160,000 watts to do less than what a $100 graphing calculator running on AAA batteries can do today. LLM power draws will drop in the future as well, and the hardware will get cheaper but right now, not charging for stuff like this is just not feasible.

 

[Funky D 272]

12 hours ago, UrgentSiesta said:

Google's also highly focused on consumer entertainment like Apple, ad revenue, etc. And their biz services divisions are doing rather well, too. But they can't seem to find their a** with both hands when it comes to gaming, though they have all the other pieces/parts for flight sim, unlike Apple.

Your whole post is pretty spot-on, but I'd argue that Google is only highly focused on advertising. They either abruptly cancel other services and products or leave them to slowly wither and die. From 2014-2018 I would have called myself a Google fan. Funny how, with the enforcement of GDPR and CCPA, all of their products just started sucking more and more. And I say that as someone heavily invested in Google's infrastructure, with multiple "smart" devices that have gotten stupider and a longtime supporter of Pixel phones. Someone at Google could create the most advanced flight sim in the world and Google would mess it up and cancel the entire project a year or two later.

Make no mistake though: Microsoft is able to take a lot of risks right now due to the cash cow that is ChatGPT/AI, not to mention Azure being up there with AWS for cloud infrastructure (Google dropped the ball on their cloud as well). With Apple's primary focus on hardware and app store revenue, and Google focused on advertising that will keep getting hit by privacy advocates in the countries that treat citizens better than corporations, neither are poised to enter the already-established gaming market (Google didn't even try with Stadia, they just threw it at the wall to see it break instead of stick).

That leaves Amazon, which is in a similar position to Microsoft due to their diversified product portfolio. And, they've tried a bit, with their MMO game New World... but if they got into the flight sim industry, it would really be coming out of left field.

That being said, other small devs could surprise us with a competitor to MSFS, but in order to get worldwide satellite data and AI for things like ATC, it would likely be an expensive monthly fee. MS can write off all those server operating costs since they literally own the servers, but a small dev wouldn't have that luxury.

Anyway, apologies for the slightly off-topic Google bashing... really wish Ballmer didn't drop the ball on Windows phone back in the day so we would have something other than iPhone and Android out there...

Edited February 24 by Funky D
Edited to replace "word not allowed"

[HighTowers 1,434]

Definitely don’t want Google doing a flight sim.  I can just see it now… a 5 sec ad pops up for ATC transmission, and 30 sec ad pop up for every takeoff and landing.   YouTube is out of control with ads now. 

[abrams_tank 7,425]

1 hour ago, Funky D said:

That being said, other small devs could surprise us with a competitor to MSFS, but in order to get worldwide satellite data and AI for things like ATC, it would likely be an expensive monthly fee. MS can write off all those server operating costs since they literally own the servers, but a small dev wouldn't have that luxury.

Yup, I absolutely agree with you.  A smaller company would pay a lot for the satellite data and for the AI technology, because they don't own it. And as you said, they would probably end up charging a monthly fee for it, which could end up being expensive.  It definitely helps being at the size of a MIcrosoft, Google, Apple, etc, because those companies own all that AI technology and have access to the satellite data, so they would make things much cheaper for a next gen flight sim.

Edited February 24 by abrams_tank

[abrams_tank 7,425]

5 hours ago, eslader said:

Your premise assumes the actual AI processing is cheap enough for a huge company like Microsoft to be able to offer it for free. It's not. ChatGPT spends a fortune to run itself. Microsoft would have to do the same. The supercomputer MS built for OpenAI has 285,000 CPUs and 10,000 GPUs.

Actually, I didn't say that Microsoft could offer it for free.  For AI voice accents, if the MSFS team determined that it's not feasible to offer it for free, they could end up charging a price like BeyondATC for the online premium voices too. But because Microsoft owns the technology, the price could end up being much lower than what BeyondATC is charging.

For ChatGPT, it's still free to use ChatGPT at the basic level. I am not aware that they are charging people to use ChatGPT at the basic level with a basic account.  But if the MSFS team determined that an AI ATC system that uses ChatGPT could not be free, they could also charge for it, but I think it would still be cheaper than what SaysIntentions.AI is charging.  Remember that Microsoft does own 49% of OpenAI so that would probably give them leverage in discussions with OpenAI to use ChatGPT's services.

For the satellite & photogrammetry streaming, Microsoft has not charged extra for that, for the 3.5 years that MSFS has been released.  So somehow, Microsoft (ie. Jorg) worked out that they didn't need to charge users extra for the satellite & photogrammetry streaming, up to this date.

In any case, if a much smaller company had to pay for the AI technology and satellite & photogrammetry data streaming, it would probably end up being more expensive because they don't own the AI technology and they don't own the satellite & photogrammetry data (and of course, owning the servers helps to lower the costs as well, as Microsoft owns Azure).

 

Edited February 24 by abrams_tank

[Greazer 1,321]

• I am also thankful that Microsoft is behind MSFS for the purposes of AI development
• Even so, I am not convinced that FS2024 will include AI ATC out of the box. Simply because there is simply too much else to do. But who knows I could be wrong.
• However, developers will have access to all the Best AI dev tools such as Copilot and Code generators etc. So I expect next version is going to be super optimized.
• I am not actually planning to do a Hardware anytime soon, as I believe my 5800x3d and 3080 Ti still can go on for quite a while yet, on optimized FS2024.
• I expect FS2024 will be more cloud based but not entirely and some addons will be picked up locally.
• I expect the scenery is going start looking better at ground level because the sim world will not always be solely for flight.
• I expect the next version is going to have multiplayer shared cockpits
• I expect the next version for Deluxe copy will have AIRAC updates and Jeppeson charts
• I expect the next version will have more Boeing aircraft out of the Box, including at least one 737 Max and a 737 (non Max)
• I expect the next version is going to have more advanced ground handling
• I expect the next version is going to run very well on PC and XBox from day 1.
• I expect the next version is going to have much improved flight modeling / physics

[Lucky38i 1,095]

1 hour ago, Greazer said:

I expect the next version is going to run very well on PC and XBox from day 1.

Good one