Comment by adventured | Hacker News Reader

Comment by adventured | Hacker News Reader

Parent Full thread

It's a complete bifurcation of the Web's technology underpinnings. That will keep getting more dramatic. Developers and their pay will bifurcate similarly more and more as time goes on.

AWS has 150 services or so now. How many of those does a 12 employee small business need? None most likely. That business's site gets maybe 1,000 or 3,000 or 10,000 hits per month. It's an absurdity to throw all the new tools, libraries, cloud tech etc. at that. Most of the Web looks like that small site, most sites on the Web get very little traffic. So little that you could run it on a $5 Digital Ocean instance and have over 95% of the droplet's use capacity to spare.

Does my insurance agent, who operates his own small business, need machine learning? No. And to the extent he does, it'll be built into the tools provided by various insurance carriers, entirely hidden from his daily business operation activities.

So who actually needs all of these tools? You nailed it, it's almost entirely isolated to the 0.1% group, the Facebooks, et al.

> Does my insurance agent, who operates his own small business, need machine learning?

I have a friend who has his own navy. It's a small navy. He does like boats.

Once he had a men's yacht trip. There I met a guy and his son who ran an insurance agency.

Well, AI or not, one of the keys to his insurance business success was his ability to separate good customers from risky ones. Basically he knew nearly everyone in town, or at least every one of his customers, down to what they ate for breakfast or some such. That knowledge helped him get a good loss ratio and made good bucks for his business.

So, maybe what he did was not AI but real I!!

I'm sure that if someone had an AI tool that would help his loss ratio, then he'd us it!

For a second example, my father in law was on the local bank board. So, a guy in the community applied for a home loan. Well, at that bank, the loan application went to the BoD of the bank -- no joke.

Then at the BoD meeting, everything about the applicant got exposed and discussed right down to the details of his lifestyle. So, again, if some AI application could help that bank make better loan decisions, then likely they'd use it.

Lesson: That local insurance agent and that local bank both were really big on gathering, analyzing, and using all the data they could get. If AI could help, then they'd use it.

Well, it would probably be built into the insurance carriers' web stack. Dev teams need to keep up with the latest tech so they don't lose talent and can justify ever increasing salaries. So I think that's part of the reason you're seeing things move towards client side.

And you're right that most sites don't see traffic numbers to justify exotic AWS setups. I personally think most web businesses grow pretty linearly, so the focus on "overnight" scaling problems is more of a sales tactic than it is reality.

I do think you're underestimating the amount of companies that do need these tools, though. It's not as binary as "insurance broker" and "Facebook". Logistics companies, payment processors, restaurant groups, grocery stores, etc. all have pretty complex backend services and some of these tools are critical to even compete in their industries.

“Does my insurance agent, who operates his own small business, need machine learning? No”

If he has a fixed mindset you are right. But if he has a growth mindset and has the intellectual curiosity to learn what ML is and if he can apply it to his business and finds out that he could maybe he can increase his business 10 fold. Even if you take a stance that he is not interested in growing just the fact that living in this age and time gives him(or his kids) the option to learn/play with powerful tools at a click of a button is worth appreciating on so many levels

A major difference here is simple economies of scale.

E.g. a particular technology feature might cost $100k to implement in your business and get you a 10% revenue increase. For a large business, doing so is a no-brainer; for a one-person business doing the same thing is clearly a waste of money and effort. And there are many things like that, causing the bifurcation between the needs and common practices in large companies vs small.

That’s exactly what today’s cloud technologies bring to any small business. You don’t have to invest $Y to know if you can increase profits by x%. You can invest $Y/10 (even less in some cases) to figure out if this works.

If you zoom out and think about it for a second what most technology is doing today is making a small guy look big

Reply on news.ycombinator.com