Hacker News Reader

Comment by bmelton | Hacker News Reader

I think it's fair to say that any of Google's individual technologies can be replicated in fairly short order, but the problem is that the combined effect of Google's collective suite of tools is powerful.

Yes, I can build a Gmail clone in a few months. No, I can't get Android Assistant to automatically read invitations from it, add them to my calendar, generate maps to the event on the fly, and tell me when I need to leave to avoid being late to get there.

So yeah, it's a moat, and it's a pretty damned big one.

View on HN

Google also has a lot of assets that are just too labor-intensive to replicate with any efficiency. Apple found that out when they tried to compete in Maps.

And for all that Google indexes and initiates from their end, every SEO worker on earth also makes sure their client's sites are properly formatted and submitted. That kind of buy-in is reserved for companies that can demonstrate you'll get a return on your investment.

> I think it's fair to say that any of Google's individual technologies can be replicated in fairly short order, but the problem is that the combined effect of Google's collective suite of tools is powerful.

What? That makes no sense to me. There have been thousands of some of the smartest/highly skilled people (payed some of the highest salaries) in the industry building systems for decades, that's Google infrastructure today. How do you expect to replicate that quickly without similar investment in time and money?

The hard to surmount "moat" is inherent to the tech and its demands. Just take the amount of dark fiber they have or the highly customized datacenter hardware and software stack and the sheer amount of them. Sure, you can replace all that with a smaller set of VMs using someone's "cloud services" but your margins, power use efficiency and scale will be much worse.

That reminds me of a discussion with a Linux filesystem engineer: building a POSIX filesystem is easy, building a thread-safe highly scalable fast and efficient POSIX filesystem is hard. You can have a single operating systems student build a simple POSIX filesystem in one day but it takes year-hours of multiple highly experienced system engineers to get a production level filesystem.

I didn't mean to trivialize the effort, or diminish the amount of work. I just mean that most of the hard work has been done by being thought of already.

Leveraging spam detection algorithms across all accounts is a brilliant idea, and a paradigm shift from what people were doing. A genius thought of it. Relatively speaking, it doesn't take a genius to replicate.

Also worth remembering that AJAX/XHR technologies were very new when Gmail was being built. Fleshing out the interactivity on the site was a herculean feat relative to the amount of effort it would take to rebuild the thing today in React, Vue, or Angular.

The groundwork they laid in building Gmail is now available to the commonfolk, so yes, replicating the effort today would take considerably less time than pioneering the effort then, and I don't think that's pejorative in any way.

Reply on news.ycombinator.com