For example, can you create a product that will reach 5K MRR ?
825 karma · joined August 11, 2017
For example, can you create a product that will reach 5K MRR ?
So the problem with AV is not the AV but the HV. If AV cannot communicate with other AV, or at least trust that other cars will not behave irrationally, they will never be able to self drive.
With mixed AV/HV those hard problems cannot be solved.
Note that this true for any kind of robot.
Look at amazon warehouse robots, they are operational in their own part of the warehouse, while humans are operational in other parts. This is one of the reason that we do not have home robots.
Another option is a personalize self driving AI model per driver. I.e. most driving occur in repeated manner (same time and same routes), so training a self driving models for each driver routes, might solve most of the issues (I.e. let each driver train his/her car model).
However, Tesla business model rely on a driverless taxi which suppose to support all routes in all times.
If you take VC, than there is always an existential risk, since you are relying on new money being available for the next round.
Also, you diversify in the stock market mainly due to information uncertainty. I.e. you do not have any control on different risks (economy, company corruption, etc.).
However, in a startup you have much more control on the company future (again, if you DONT take VC money).
If you take VC money, than the company is basically a privately traded company, and you assume all the risks of a public company.
Note that the earning are protected mainly because you invested the time , and not because of some union.
The Achilles heel of those professions, is that the body of knowledge is constant, with very long half life. But this also make it ready for automation.
With AI, the destruction of those professions will be quick, as they will get destroy from the bottom. I.e. Even if AI make existing lawyers twice as productive, it would mean that you would need half the lawyers. If AI will make them 4x productive, you would need 25% of lawyers, etc.
In addition, programmers have many more domains that they can switch into, however, once the AI lawyer or AI doctor will rise, it would practically replace 90% of the lawyers/doctors.
2. When I refer to community first, I mean that the product was rushed to market, half baked, and than was iterated in the market. I think that Google worked with redhat on the initial API (I.e. deployment object came from redhat)
Regarding AWS, AWS does not have 5% of the global IT market, of course it have 40% of the global cloud market, but not the IT market. Also, the usage of AWS or any other cloud for that matter, is very skewed, I.e. 63% percent of the customer have one S3 bucket, or one EC2 node.
My point about open stack is reflecting your point about VMWARE / IBM. I.e. kubernetes will end up maintained by companies that sell multi cloud/on prem solution.
Overall, I believe that kubernetes is the only way the IT world can decrease the power of the big 3 cloud providers(I cannot see any other option at this scale), hence usage should be encouraged.
1. Of course kubernetes was developed to get customers to GKE and for profit reasons. 99% of the open source developers are paid programmers. Hence there is an alternative cost to their time. I am not sure why this is wrong. At least they did not fork an open source project and offered it without credit to the original companies, like other cloud providers.
2. The way google develop open source products is community first / tech later. But this is stated. For example, kubeflow - it was started as a community project, got the requirements/community in place first, and than was redesigned I think 2-3 times. The one thing about google / GCP is that they are doing the initial R&D, but fail to follow - I.e. they expect the community to pick up. I assume that this can be attributed to the internal promotion cycle inside google (where engineers are expected to rotate), vs tech reason.
3. Of course kubernetes will be maintained by on perm companies (e.g. VMWARE/IBM). The predecessor to kubernetes was open stack, which was also developed by on perm hardware companies (e.g. CISCO / HP). You still need money to pay engineers.
Moreover, I think that you are missing the key attributes of kubernetes, which morphed from its original goals. Kubernetes is about 3 things:
1. Scalability - you can run it on you laptop or IOT device up to the cloud (e.g. k3s).
2) Distribution - Kubernetes as a platform is permission-less and widely spread. I.e. If you want to extend it or develop on it, you do not need anyone permission, and you can assume that it would exist on your customer cluster. It is my belief that AWS have about 5% of IT workload, and is hitting a wall (see the latest sessions on AWS conference, which is catering to more legacy companies).
3) Automatic operation - Kubernetes control loop allow machines to manage machines (both software and hardware). I.e the carrot of the cloud is offloading management of your infra. The stick is that you pay 20x more on your hardware. Kubernetes give you another way - which is encapsulate the IT knowledge in a automatic robots - called operators, and let those manage your application. The problem is that this feature is not widely known. (and it is my goal to change that).
Also, since everything is hidden, when a time come that you need to debug, it is much harder.
There is however an ongoing budget increase in marketing to programmer/manager telling them that programming is getting easier.
you need to practice (a lot) both inside and outside of your job. Read books about programming (not about specific tech) and make sure that you understand what is written, and than try to apply the patterns in your regular work.
Forget about promotion. A promotion is some external reward that is given for reasons outside of your control.
My point is that the by-product of big tech is a useful set of tools which would not exist without substantial R&D investment subsidized by revenues from ads.
Also, I am looking at reality vs what should be. If given an option, consumers would not pay, not even 0.99. This is reality. I.e. consumer are willing to give away their privacy for 0$ and free tools.
Most of the open source tools today (e.g. kubernetes, tensorflow, pytorch) are not coming from academia , but from big tech companies.
The problem is not those companies, but rather the consumers do not want to pay for anything online. Hence, the need for advertising, privacy issues, etc.
I.e. as soon as it starts replacing humans, it will not have enough human generated training data, since all of programming will be done by models like himself.
Second, alphacode was specifically trained for competitive programming :
1. short programs. 2. Each program has 100's of human generated solutions.
However, commercial program are:
1. long. 2. Have no predefined answer or even correct answer. 3. Need to use/reuse a lot of legacy code.
The complexity of a software system is due to interaction between its components. A professional developer know how to make it N instead of N2.
Note that most of the big software system that you know, were written by a single/two developers - C++/ Java / Unix / Linux / Riser FS / Kubernetes.
This can be evident by the power law distribution of github commits for every popular github project.
I.e. a lot of people can understand 10-20 lines of code, less can 100 LOC, less can 1000 LOC, maybe top 10% can understand 10K, 1% - 100K.
So no code can work, just that the size limit of the program will be low (10-100 LOC) .
However, most program start at low LOC, but ends up with high LOC, where the low code tools breaks.
Hence, the low code tools basically guarantee the future of the elite geeks.
The fact that he was accepted to an incubator should be used as validation of the idea.
Most markets can support 10's of startups.
A war require the players to show their cards, which is in nobody interest. Also, there is a chance they you might lose.
But "Crashing" is point of view. I.e. if you short the market (a seller), the market is actually up. I.e. you should not try to time the market but time the DIRECTION of the market.
Also, the Fed is not hawkish, if it could it would not raise rate.
IBM is a sale driven organization. I.e. sale first - tech after. so it cannot be judged as an high tech company, I.e. on the tech.
A point of reference is SUN - which was tech first - sale after.
Another point of reference is HP. HP was tech first, but turned into sales first.
I.e. VR is the leading indicator to the metaverse (somewhat like the iphone = Without proper hardware, there will be no software).