GmVault: Gmail account recovery mutt-gmail
Hope you & family are well. Wendy would prefer I didn't ask you this question 😉, but I'm hoping you have time for a chat (contact info at the bottom, call anytime) It would be easier to explain the jist of the situation (just tell you what you need to know) over a phone call rather than show you the 9 page research document related to this (unless that's how you prefer to consume something). Perhaps the ideas are bad and totally infeasible for some reason I don't know. I hope you know the answer or will be equally curiously amused by the question and find the answer in your network.
Phone number below. Definitely not urgent. Happy to do video on whatsapp or zoom as well, based on your availability or call me from the car at your convenience.
background for your reference, not important.
The nature of the idea/apparatus is a front-side voltage regular on a solar panel (on the 12v side), along with a storage system to supply the inverter.
I think I could boost solar output at least ~5% for less than <$100 per system AND more importantly store power "more intelligently" than we currently are in the grid to better utilize it.
In my head it'd be something that could be easily installed by any qualified solar installer or sparky? in < 30 minutes to any inverter (or a feature that could be added to existing inverters that have WIFI as a software update, or run directly through AEMO). It's one of those "why doesn't this exist yet questions" .. maybe it's a stupid idea. It's probably a whole series of stupid ideas. But I believe the reason it doesn't exist (Occams razor) Nobody has attempted or successfully built one yet.
"Rockets Corollary" to that which is "why not try, it's probably easier today, and could make an awesome future." Whereas Wendy is "Ground" that says "that'll never work", leave it alone as she had hoped I would do with the grid idea.
Why I care/My situation: Trying to get feedback / awareness. I'm presently building a system called b00t http://github.com/elasticdotventures/_b00t_
That is a deployment tool for my various bits of code I've been building over the last 5 years, it will go "online" after a series of massive AI training "learning" sessions, that system initially will be a really smart phone system (called AIIA), and it's important to realize that AIIA is the call center for GrowPotBot.
It's a common pattern for technology companies to have a commercial or individually owned system built on top of an open-framework everybody can use.
AIIA will later this year/early next be able to answer the phone, access catalog data from a website, answer a variety of questions that would normally be done by an employee. The learning aspect of the system (the follow up call to the customer to make it more intelligent) will be something unreal. Again, I expect somebody to release something like what I'm suggesting soon, but it's not well suited for an academic paper so it also probably gets overlooked by a lot of smart people. This product is too big for a small research team, and while it's probably being worked on at some level inside a big company like facebook it's designed/optimized "trained" on their business cases -- not mine.
I've seen/know (approximately) how to make software that writes software (by watching tutorials, looking at other software) and then to take that knowledge and build other software (to learn programming, or other skills it can interface with such as phones, robotics, etc).
These are the "sci-fi" types of systems which can (in theory) design/simulate its own electronics (maybe even design & order it's own parts from the Internet someday). We're living in a present reality of $40 32-bit 3d printed robot arms. WHY are people still working? So the system can learn tasks, archive the skills, then learn more skills (but not forgetting any, ever).
Building the voice interfaces and everything at the moment and looking at what my power requirements are is scary when I think about what it can grow into at scale, i.e. I want it to be running in all the clouds and access all those services and so I came up with the idea of making the software be able to pick up and move across clouds (so, in this example, the software can migrate around the world following the sun "cheap power", if I could buy power cheaper then, that would be better) .. does that make sense?
So the data centers should raising/lowering their prices with the availability of renewals. In about 3 months from now there's going to be a massive shift in the cryptocurrency Etherneum where globally it creates a surplus of cheap & idle AI chips (Ethereum will become 99.9% more power efficient than BitCoin -- this is already scheduled upgrade [it can't be reversed]). I can't tell you how the market is going to behave to that information (it will probably be oblivious), but that those miners will move their usage to other dirty coins which is exactly what you (as the grid operator) don't want (for a variety of reasons I can explain in models if you care)
BUT I'm suggesting that there are national energy policies which would benefit Australians AND collect foreign investment with even just a fraction of awareness by the government how to properly do that (and also make a HUGE PR win, for the environment, all while actually doing little/nothing, besides influencing which crypto-coins people invest in so they are more energy efficient -- which is a very GREEN and GOOD thing to do).
I would also propose a dirty surcharge, and PoW (Dirty coins) like Bitcoin transactions incur an "power surcharge" .. sort of like how tobacco is absurdly marked up here. But then you're influencing the market, how "dirty" the crypto-coin is, it's like an environmental surcharge, and it's something that AEMO should *(imho) consider within it's scope and charter. You're telling crypto-operators that efficiency of electricity matters, so like have the government come out "Pro/Neutral Cryptocurrency, Anti-DirtyPowerUsingMakesEverybodiesElectricyMoreExpensive-Coin" (i hope that makes sense)
Even if you simply collect the money to pay for Australians energy bills (redistribute it), you're going to be encouraging a good responsible Australian energy policy. I'm suggesting that the policy above could be implemented at crypto-wallet exchange, as a regulatory compliance in a few years (but it would have an impact on how operators plan/think today). Australia suddenly becomes very desirable if other countries adopt similar policies (since Australia has cheap power when more than half of the world doesn't)
Azure should be able to do this (honestly, the government should ask them nicely do this, then make then do it -- which is bill for actual power consumption, at a data-center level), i.e. their rate needs to pass through the electricity consumed, on a per-customer basis. I have a few other thoughts on Australian Crypto-currency policy (and how it impacts energy usage) that might interest you. Australia is a great place to farm power, it's sunny here when it's dark on the other half.
Anyway my "cheap electric" adventure doesn't work yet, because I need to either hack into AEMO or Amber to get the utility rates, because there are no public facing APIs for that information, which is probably why the earlier situation has occurred (does that make sense?) so if AEMO offered a public API then the businesses/consumers could make much smarter decisions about when to their power equipment (this is important if they start paying for peaks, which would require some commercial offering like Amber Electric -- does that even exist?)
In my opinion if AEMO is setting the market rates, then it's no different than the LIBOR rate, or any other global indexable number, it should be as a publicly accessible (or free registration for citizens) API. Can you ask your IT department why I can't access that?
The good news for AEMO cyber/sec, is this time they dodged a bullet, of not having to face down my "hacker AI" -- it's easier to reverse-engineer/drive/scrape Amber Electric's app and publishing their private API is what I'm going to do.
I need to stress, some of these projects are really weird (like this is my reality), it's software that types back -- this isn't software like you've seen before, it can learn interfaces and a variety of skills such as typing, moving the mouse, filing, prioritizing, re-prioritizing, etc. (but it has all the skills of the entire organization if you only used open-source software). It could also interface with a variety of non-cyber species using brain machine interfaces - in that sense I'm design a digital 3rd hemisphere for my own brain (someday, hopefully) - but immediately just something that I can talk to and can be trained to print itself a tool. It's a bizarre world when you consider you can print circuit paper on an inkjet, so it's not as unrealistic or crackpot as it might sound.
I have NO IDEA how expensive it will be to operate (yet), rather that it will almost certainly someday be much cheaper than any human after rounds of iteration, it also won't quit or take sick days or sleep. Anyway to reduce that operating cost, which is continuous (in that example)
THAT is why it is not totally finished yet -- I'm not sure if that's something "new", but I suspect it'll be a future common use case. (The cool thing about this project that I'm building is once it's finished learning, then it'll use very very very little resources, but it's the learning that is super power & cpu intensive!! so this is only a near term issue). I'm recruiting other people so we can do distributive training (in various IT automation roles). At some point, if the AI can hack into AEMO (or any government system, which it's explicitly not allowed to do unauthorized/antisocial behavior)
I'm planning to teach it stuff like video editing, CAD, etc. (all using open source, free tools, and hopefully show it how to build it's own tools) -- so it's probably easiest to think the final product as a "file transformer" so you can take any input (even a voice description), any artifacts to render "files" of a scenario (using voice and google assistant) and it produces the necessary files for the project, runs the project, manages the project, cleans up the project.
I call it a "svelte team" (its you, and whoever you give access to), and it's built at the level of engineering, so it's the kind of thing you'd want to securely run a personal brain machine interface, secure posture, etc. but between now and then, it's a lot of technical plumbing to get it online.
This is the deployment system growpotbot, it's designed to create and operate ~9 billion home gardens (among other things I'd like to see in the world) -- hopefully on Day #1, BUT it also needs thousands of nodes operating on several generations to learn, or a lot of nodes and fewer generations (if that makes sense), this thing is totally overbuilt because it's managing so many different sensors in many different environments) concurrently.
In my opinion that might put a dent in climate change, and hopefully solve hunger, mosquitos are next (seriously) .. I've sufficient runway under my own steam to build this (i think) at this point, I feel like somebody smarter than me is going to release one of these ANY DAY now, but I'm usually 3-5 year ahead of the market. I worry that THAT system (*possibly from the military) may not be as nice as mine so I work on this A LOT, because as an American I'm supposed to be ready to repel the US government at any moment, it's my civic duty.
The point is .. this other product/idea is absolutely something I expect my system could design, build, blabla .. ( you/i/any team of engineering students could also build now, but I think I can design a system that could design a system like that better than them -- if that makes sense, so I want to keep working on that). Maybe I'm not as smart as I think I am too, but I know I'm sufficiently stubborn, so while there is breath in my body I'll keep working on this (as I have for the past 30 years of my career). I'm perfectly happy to give it to a team now and let them see if they can build and launch something faster than my system can (using the best practices they know how).
But I don't want to put the capital/time into a system to build this system if there is some reason the product won't work (because I have other products, but the more scenarios I can feed it, the more it can learn, and it has a feature called "transfer learning", so it can (for example) learn new video games from a very small amount of time playing with a joystick. I haven't applied this to product design or other fields yet, but I expect to be able to do that sooner than you'd probably expect me to be able to, but definitely not going live tomorrow either .. but important prototypes, progress towards a goal that the 12 year old version of me who started dreaming about this would say are "wicked cool" (building systems for self driving cars and such) so who are you to tell me what I can't do now .. I'll have a hacking AI definitely by the end of the decade probably much much sooner than you'd expect.
If you feel there is anybody who should know that, who could benefit from that information, feel free to share it with them. I'm publishing a lot of my code as I go, and even those pieces (I think) will be very disruptive for a variety of industries like it/automation.
🚀 Brian Horakh +1-760-786-4212 (AEST office hours, US afternoons are best) +61 - 422 863 313 Or let's connect via something other than email .... 🌱 Skype: brian.horakh || Wechat: elastic_ventures || Twitter: @brianhorakh || LinkedIn Whatsapp 🇺🇸: +1-347-708-3850 || IG: @rocket.wandering
IMPORTANT
Communications contained within this message are private and for the intended recipient and should be treated as confidential by both parties unless otherwise specified.
$=_=>$=${$};$();$()