Hacker News Reader

Comment by paxys | Hacker News Reader

This phenomenon is well known in tech. "The product doesn't work as well as expected. We can painstakingly investigate all the different issues one by one, root cause them, and put in the effort to fix them. Or we could just throw out the whole codebase and start again from scratch. That's sure to be more effective right?" Spoiler – it isn't. Musk just threw out a thousand+ years of combined experience in the area, and the new team is going to come in and repeat years' worth of the same mistakes in order to realize why things are the way they are.

View on HN

Throw out and rewrite can work. If you know that the old code base is beyond repair, you know how to rewrite it and are able to estimate it correctly.

What often happens is a rewrite that keeps the same mistakes as the original product (either organizational, specification wise or technical).

I have no idea about management, I just constantly see restructurings of departments going wrong.

The likelihood of doing that successfully goes up the more you know about the problems with the old thing and the more people who were involved with creating the old thing are still around.

Just firing everyone and starting over from scratch is never going to result in an improvement. Even if the problems were because the original team was just incompetent (which it rarely is), you aren't going to get a better team the next time because if you knew how to hire a competent team you wouldn't have hired an incompetent team the first time.

People aren't code. Sometimes teams get a culture that you want to change and it's generally extremely difficult, if not impossible to change team cultures incrementally because a) they always hire people that have the same culture as them, b) they're used to the existing culture, and c) people hate being told they have to change.

So I think it may not be as crazy as you make out. On the other hand were they really underperforming? Isn't Tesla's charging network still world leading?

If there is a bad culture then work to fix it. If there are underperforming employees then root them out. Like you said, it's not like 500 totally incompetent people built out the largest and most powerful EV charging infrastructure on the planet in a matter of a decade.

Not all processes are reversible, and not all problems are fixable. When there are solutions, that doesn't mean the ROI is there.

That's my point. Sometimes the best way to fix bad culture is to start from scratch.

How much of that is the result of the individuals on the team themselves and how much is that the result of the organizational pressures, constraints, and expectations they're working under?

I suspect that just replacing the individual people would result in more or less the same unhealthy culture unless you also took steps to change the rest of the organizational context.

Do you realize who you are talking about? This is a guy that has done this multiple times in the past and somehow always has an endless number of bodies to burn through to accomplish the task.

He did this with Starlink. One day he wasn't happy with how fast Starlink was progressing and so he went up to Seattle, fired the entire team and started over. That led to the Starlink we have now. I have spoken directly with employees there and they totally believe in the mission more than anything and they want to only work with other A players. Again, I am amazed that Musk manages to keep attracting top tier people to kill themselves over his vision but i'm not an A player so what do I know.

I disagree. Rewrites can often get rid of more debt than they generate. They are always scary for psychological reasons, nevertheless.

I'll add my two cents here. A lot of the work that goes into a codebase doesn't necessarily appear in the final product. Sometimes code is hastily written to get the system working, and then it has to be rewritten to be faster or more stable or what have you. A rewrite can just reimplement the final code. Sometimes the company changes direction (think of the stereotypical aimless startup that pivots every Tuesday) and the code solves a problem that the company is no longer interested in. A rewrite can just ignore that code. Some of this has to do with technical debt, but it's also due to the fact that it's easier to do something when you know exactly what you're trying to do.

It's like the difference between between reproducing a proof from a math textbook and actually proving a new theorem yourself. The latter takes a lot more time and effort, even though the amount of work in both cases looks the same from the outside.

you mean how twitter was gonna break after so many fires but it didnt? yeah he probably knows a thing or two

P U S S Y I N B I O

But seriously, how can you not go to Twitter and see that it has become totally enshittified. The real lesson learned there is that large tech infrastructures are actually stable enough to survive a lot of BS without collapsing. Doesn't make the products pleasant to use, though.

The Twitter acquisition is like paying 500k for a house that’s worth 300k, turning around and gutting it so it’s only worth 175k, then claiming victory because “hey at least it didn’t burn down”.

I think it’s hard to see what has happened to Twitter as anything but utterly breaking the business.

I wouldn't call X-Elon-Twitter a success. It was more like not a complete failure.

The point is, there isn't literally a smoking crater where Twitter HQ used to be, therefore all criticism of the acquisition is invalid.

Reply on news.ycombinator.com