I just got a voting system idea (which might or might not be an already existing system) after watching two videos about voting.  
Disclaimer: The question(s) will be at the end, but the main reason I am asking this is because I would like to hear some feedback about this idea and learn new stuff as I am fairly bad at politics and I do not know too many details in this field.
The idea would be that every citizen becomes eligible to state their preference, their supported politician or party (or both) at a given age - like in other voting systems. When anyone states their preference, that supported person becomes their preference until they change that to someone else or that political instance gets retired/disbanded.
This means that at any given moment there is a clear overview about the state of support of the political parties. This way there is no need to hold elections periodically as the support distribution would change rather quickly (more about this later). Politicians or politicians from parties would be elected as soon as the support percentage changes with a few practical restrictions:
Every winning person should gain a noticeable margin (e.g. +5% over the old elected president/prime minister/etc.) or the winning person should keep its first place for a specific amount of time (e.g. +0.1-4.9% for 2 months). This way there would be re-elections less frequently and less money would be consumed.
Elected persons would remain in their position for at least a given amount of time (e.g. 2 weeks), so they could prove their abilities if some initial citizen anger would try to remove them.
The main idea of this voting system would be the dynamicity: if an elected person proves to be incompetent then it would be replaced in a more effective way. Sometimes (when some bad thing is done by the government) people would be voting in larger chunks like in currently held elections, which could backfire with huge waiting queues, but this might be solved with local regulations (like opening up voting booths on weekends if more and more people start changing their supported person).
The current paper voting systems have good advantages, like anonymity and trust.  This new system would probably be less secure and would probably also have less anonymity over old systems. Even though the support percentages would change rather quickly, practicality and the disadvantages mentioned earlier would suggest that these changes should be applied with some periodicity and some batching. My idea would be, that every given period (e.g. week) the new votes would be sent to a central place, if there is a higher count than some number defined locally (e.g. for 1000 inhabitants there should be at least 10 votes on a weekend, so those votes could be sent).
Some concerns of mine about this voting system:
Less people would vote than in current voting systems, because there would not be a specific day for voting. Maybe people who are not into politics and politics related stuff would ignore this system.
The issue mentioned earlier: lots of people would start to change their supported politician or party when some bad decision is made, overcrowding the local permanent voting boots.
The previously mentioned permanent voting boots could be a higher expense than periodical elections.
Referendums and setting up ministries might not be that straightforward, I guess?
It's most likely less secure than a classic periodical voting system and the issue is greater on smaller governing instances, like small villages.
What I think are the positive traits of this idea but I am happy to be proven wrong:
Bad decisions would result in fast change of leadership.
As permanent voting booths would waste some money, we also could save a lot on periodical money waste (like campaigns)
There would always be an overview about the distribution of support between different political instances.
Electronic voting based on secure identities (and here I am taking the idea of SQRL as someone interested in this new concept ) could partially solve the problem of anonymity and trust but only if used with caution and regulated by government.
So, getting to the end finally... I would like to get some feedback, insights about this idea. I want to be proven wrong or be supported and I obviously want to learn about politics in the meantime.
And thus, the main question is: could this idea possibly enhance most voting systems or it surely would be a total backfire?
I know that the question might be broad, but I am willing to accept answers if they do provide answer to most of my concerns. Also, here are the sources mentioned throughout the question: