It’s more like buying a shopping district where there are public areas where you like to hang out, piss off all the shop owners, who are currently paying all the costs of upkeep, tell all the movie stars who frequent the place and who draw in the punters, that you’re going to charge them, and the punters a fee for getting in, because that will keep out the riffraff, but they’ll now be able to use cafe tables without buying anything … Okay, this is getting away from me. But the most concrete thing Musk has done, which one can put some numbers on, is promoting a paid tier as an essential change. The cost of this appears pulled out of his butt and it will presumably only be attractive to people who spend a lot of time on Twitter. Those people are also the ones that supply the most ad views, and there have been back of the envelope calculations showing that if you show those people half as many ads, that costs Twitter more loss in advertising money than their fee brings in. Not to mention no one has really managed to introduce fees to online media and keep mass appeal.
Yes there is a lot of hyperbole, but there are tangible negative effects of several of Musk’s choices so far.
All that stuff (contained in what I called “engineering judgment”) only gets you to an X% confidence level, where X% is strictly less than 100%. Since 100% is impossible, what threshold do you consider acceptable? And what is your justification for said threshold? There is a cost to increase your confidence level, as well as a cost to false positives (services that appear to be in use but are actually dead).
If you are too sloppy, you will have a bunch of false negatives and few false positives. If you are too careful, you will have a bunch of false positives and few false negatives. Neither extreme is good. You want to be somewhere in the middle. False negatives are likely more costly than false positives, but not infinitely so. So pick a ratio, say that you will accept 1 false negative for every 10 real examples. If you are not experiencing enough false negatives, then you are not being aggressive enough.
Sometimes a “service” itself isn’t even a well-defined thing. At work, one of the internal services changed the configuration on their web server and the format of the directly listings changed a bit. Same information, no big deal, right? Except that other parties (including me!) had been depending on scraping the page for listings and the new format broke the scripts. Ok, shame on us (though we did ask for a proper API and didn’t get it). I don’t blame them at all for not anticipating this; I mean, they could have spammed the entire company to see if anyone, anywhere might have been depending on this particular thing, but that would have been a poor use of resources. Better for them to just make the change and see what happens.
Starting with the premises, I don’t believe that Elon Musk has no business acumen at all. He knows a lot about subsidies, flattering politicians, getting handouts, self-promotion and things like that. Some of it is useful, at least for a while. He also knows how to motivate his staff, and be it with duress.
And I don’t think that he is the richest person in the world.
As an aside: can you still buy Teslas with bitcoins?
In just about every endeavor of his, he is much less good about these things than his competition. In 2014, both Boeing and SpaceX were awarded funds for NASA’s Commercial Crew program. Boeing got $4.2B, while SpaceX got $2.6B. Same requirements, but Boeing got 60% more money.
In 2019, Boeing got another $290M despite this being a fixed-price contract. SpaceX did not get any bonus.
Today, SpaceX has sent 22 NASA astronauts into orbit over 6 flights. Boeing has sent zero people into orbit.
Which one of these sounds more like subsidies and handouts?
I know you like him, and you know (I guess) that I don’t. Could bombard you with articles, mostly in German, about his Tesla factory (Megafactory? Hyperfactory? I forgot, something along those lines) south of Berlin, how he got subsidies for promises unkept, violated the water use laws, the zoning laws, employment legislation, environemental legislation on so on and so forth. No point in arguing that Boeing is even worse, because I am not talking about Boeing. And yes, they are bad too, but so is Airbus, for whataboutisms sake got plenty of subsidies too, and they both got our respective countries into a fruit- and pointless trade war. Still, my point is simple: Elon Musk would not be where he is now without all the government help and subsidies he got. And now he is against government because censure! libertarianism! and because he is an ungrateful brat.
Feel free. But my point is not that his endeavors are not subsidized. Every major business gets subsidies, whether in the form of direct payments, tax breaks, or simply getting sweetheart contracts. And this is doubly true in the case of well-established industries like automotive or aerospace.
It is not whataboutism to point out that all of them are being subsidized. Nor that we seem to get much more value from Tesla’s or SpaceX’s subsidies than we do from the other businesses.
After all, the whole point of subsidies is to encourage investment in a certain direction, say for electric vehicles. Of course those subsidies will be taken advantage of. The reason that subsidies has a bad reputation though is that many businesses simply do not follow through.
Early on (2009), Tesla got a ~$500M loan from the US gov (DOE, IIRC). Ford got a $4.2B loan through the exact same program (intended to push EVs). Tesla paid back their loan early, and we got several good vehicles as a result. Ford squandered the money and only very recently has a reasonable EV product line.
You mentioned the “Giga Berlin” factory. I’m aware that there have been some issues, water use, etc. But that sounds like the kind of noise that follows around any major project, which can be amplified or not depending on which news sources you follow, and which usually gets resolved quietly. However, the factory is now producing 2000 cars/week and is on track to producing 5k/wk early next year. Perhaps this is later than expected, but are major projects really ever on time?
Ultimately, I like to see results. EVs actually being manufactured, rockets sent to orbit, and so on. I am not too concerned if there was a subsidy if it actually resulted in a positive outcome. I do not like it when incompetence is rewarded. And scale matters, too. Company B should not get twice the money as company A if they do the same thing.
Whatever Musk’s motivations may have been in buying Twitter, they have (so far) backfired very badly. The major motivation for the philanthropy of the 19th century robber barons like Carnegie and Rockefeller was to try to gain social respect instead of being seen as the self-serving ruthless thieves that they were. If Musk had any such thoughts about improving his image, boy did that ever blow up in his face! Many people who previously saw him as the superstar innovator who successful ran both Tesla and SpaceX have now had a front-row seat to a spectacular and very public display of comical incompetence. Prior to his involvement with Twitter, I doubt we would have had a thread asking how someone so incompetent could possibly have become so rich.
I don’t have feelings on Musk one way or the other. But people seem to be conflating whether Musk has “business acumen” with whether he is a “good citizen of the planet”. One can run a profitable business and still be a massive a-hole.
As for whether he is the wealthiest person in the world, Forbes seems to think he is (at least as of the articles date of publishing). Whether he got there from government subsidies or started with a bit of family wealth is irrelevant IMHO. The government presumably didn’t just arbitrarily decide to make Elon Musk a billionaire. And his family was “well off”, not “wealthiest people on the planet”. Securing financing for your business (whether it’s the government, banks or private investors) is part of running a business. Looking at the market value of his companies, it would seem he produced a pretty good return on those investments.
Musk’s motivation wasn’t (IMO; he doesn’t tell me his plans) exactly philanthropic or reputation-burnishing as e.g. Carnegie’s was. More like he believed (confusedly) that free anonymous unfettered speech was a Good Thing. So he intended to deliver this Good Thing to the masses. And maybe be hailed a heroic genius for his munificence.
In a darker formulation, he just wanted to troll the world. I am the biggest man with the biggest megaphone in all of history!
My name is Musk, tycoon of tycoons:
Look on my words, ye Mighty, and adore!
Nothing beside remains. Round the decay
Of that colossal 404, boundless and bare
The lone and level debts stretch far away.
Regarding this, and the “X% engineering confidence level” you mention.
First of all, screen-scraping as an application interface isn’t just asking for trouble, it basically guarantees it. Not your fault, but the fact remains.
But I strongly push back on what appears to be your suggestion that trying out changes in production is the only sure way to gauge their impacts. Does anyone seriously think that critical systems – like for example banking systems – operate that way? In the early days of the corporate intranet, as a consultant I once had to move heaven and earth to convince a bank that changing the contents of a website should not have to be subject to 17 layers of change management protocols! Can you imagine how they would regard the idea of “let’s just try it and see” to making critical changes to operational software in production?
Understanding and compartmentalizing the impact of changes is the fundamental purpose of that misunderstood but critically important concept of software architecture. On a very large scale, one might call it enterprise architecture. The fact that companies like Twitter not only failed to implement it, but apparently were clueless that such a thing even existed, is beside the point. The existence of a system architecture means that all your interfaces are well defined, and system dependencies are well documented. It’s not a naive pipe dream, it’s a scalable reality, though few organizations have had the discipline to achieve it. The CORBA architecture and IBM’s multi-platform MQSeries products were early attempts to provide actual standards and tools for achieving integrated corporate information architectures.
I worked for Network Solutions in the 90s and there were companies that built their whole business model around scraping data off our web pages. We overhauled the web site and someone pointed this out, and senior management basically said, “Tough shit.”
Although I alluded to it, perhaps I should have emphasized it more–the amount of risk one tolerates depends on the cost of a false negative. Of course a bank will put an extremely high cost on even small failures, and therefore should demand “many nines” in their desired level of confidence before making serious changes. But then, any service would not even be launched in the first place unless it could meet guarantees about reliability and years of support and so on.
But this is Twitter, not a bank. It would be utterly stupid for them to assign the cost of, say, a missing tweet as the same as a missing bank transaction. Not only is the cost of the failure negligible, but the effects are likely transient. Changes can be reversed relatively easily. And, well, social media is faddish. If a competitor site launches some new feature that requires a collection of services, they can’t go off for a year to plan a support strategy only to come back and find the fad has already died. It’s not ideal, but hardly unexpected that some of these services might end up as zombies with no one claiming ownership.
I read this article recently about the NYTimes paywall:
They were fairly careful, as one might expect given that this is how they make their money. And yet they anticipated the fact that some problems really only show up in production. So they rolled out the new service to 1% of users, then 50%, then 100%. Oops! A bug in their replacement service caused the new paywall to stop metering a certain class of users, which only appeared at the 50% level. It was easy for them to scale back and fix the bug, but the point remains: despite the diligent prep-work they did, there are always things in production that you don’t anticipate.
That’s pretty galling for someone to complain about their free data source going away. I mean, I’ve done my share of scraping in the past–either for kicks or because I didn’t have much of a choice–but I’m not gonna call up the provider to complain if the format changes. That’s just rude.
We eventually got a nice JSON-formatted listing. Something like 937 days after filing the API bug. Breaking the scraping was really a good thing since it kinda forced them to fix it properly.