The following day Fogg apologises to Aouda for bringing her with him since he now has to live in poverty and cannot support her. Aouda confesses that she loves him and asks him to marry her. As Passepartout notifies a minister, he learns that he is mistaken in the date – it is not 22 December, but instead 21 December. Because the party had travelled eastward, their days were shortened by four minutes for every degree of longitude they crossed; thus, although they had experienced the same amount of time abroad as people had experienced in London, they had seen 80 sunrises and sunsets while London had seen only 79. Passepartout informs Fogg of his mistake and Fogg hurries to the Club just in time to meet his deadline and win the wager. Having spent almost £19,000 of his travel money during the journey, he divides the remainder between Passepartout and Fix and marries Aouda.
> Pay yourself and a VC-appointed board member below market rate, hire six good engineers, lease some office space, buy equipment, pay for legal and accounting services… and presto, you’re burning through $5M a year
Would love to see the math behind this. How does this add up to $5M/year??
Right, unless you're paying the engineers $400k/yr each, I'm not quite sure how this would add up. Also, what does it mean to pay your VC-appointed board member below-market rate? How much do people pay board members, and how would this be a material line item?
If someone's making $600k at Google, how much would they expect to make at a startup? Based on what I've heard folks tend to top out around $2-300k, with the rest in options. Much of the reason people leave places like Google is to be in a more startup-y environment, and to have a chance to build something from the ground up. If they wanted to just keep making bank, they'd stay at Google.
Well, if you are leasing office space, you are likely in a high cost location like Silicon Valley, and so you are most likely also paying your engineers a lot.
If (almost) everyone is working remotely, you can probably get by without office space and also with lower salaries.
Areas like Silicon Valley are very productive for some reason. That's why people put up with the high costs.
Leasing office space in such highly productive areas might be worth it. (I don't know exactly how the mechanism works, but I'm just assuming that at least some of the companies there know what they are doing.)
Of course, the author of the original piece also seems to implicitly assume Silicon Valley?
If there's no highly productive part of the world where you want to concentrate people, you might be better off spreading out your startup and taking advantage of lower labour costs around the country and around the world.
There's lots of smart people in eg the flyover states, or in South America, who for one reason or another can't or don't want to move.
Founder of Modal here. We've spent a ton of time on this, including building our own distributed file system optimized for low-latency high-througput workloads. We don't use K8s or Docker and built our own custom infrastructure instead.
Cold starting containers quickly is a fascinating problems. We've gotten a long way but there's still a lot more to do. For GPU-based inference, starting containers isn't enough – you also need to initialize the model GPU quickly. We are working on a long list of things that will bring down cold start latency even further.
Is Modal a good solution for running fine-tuned LLMs and Whisper models? If the cold-start time is low we're more than willing to modify our code to use Modal's infra.
Happy to follow up via email but didn't see one in your profile.
Annoy author here. What you're describing is Locality Sensitive Hashing (LSH) which is something I spent a lot of time trying back in 2009-2012 but never got it working. It has some elegant theoretical properties, but empirically it always has terrible performance. The reason I don't think it works well is that data often lies near a lower-dimensionality manifold that may have some sort of shape. LSH would "waste" most splits (because it doesn't understand the data distribution) but using trees (and finding splits such that the point set is partitioned well) ends up "discovering" the data distribution better.
(but HNSW is generally much better than this tree paritioning scheme)
I don't really buy this argument that you can rent and pay $1000/month, or you can buy an equivalent home and pay $1000/month, but now you can deduct interest rate and build equity.
Landlords can also deduct their interest expenses and so they benefit from leverage too (meaning they don't want to build up equity).
In a reasonably efficient market, this means that the equivalent cost of homeownership would be higher than rent, to offset these things (the rate at which you're paying down the principal, and the deductability of interest rate expenses). This is obviously a very simplified argument, and there's a lot of other factors going into this.
> I don't really buy this argument that you can rent and pay $1000/month, or you can buy an equivalent home and pay $1000/month, but now you can deduct interest rate and build equity.
Back in 2013 I was renting an apartment for $1600/month, but my dad made me do the math and I ended up buying a condo about 10 minutes walk away, larger and far higher quality everything and the mortgage was only around $800/month. Even with the monthly assessment fee (for the management company to handle the property and facilities) it was lower than what I was renting for. Just had to get over the down-payment hump.
I don’t think this is right. An efficient market the costs are equal. I think you meant this - if you fully load the economic story the cost of ownership and renting should be equal. It’s true though that markets are rarely ideally priced. The ideal price is a mean reversion target but can randomly diverge due to a ton of other idiosyncratic factors. In fact housing prices and rents are rarely in balance. They just tend towards each other.
Sorry to hijack the thread here. Why is your prediction that IBM gives up "hybrid multi-cloud"? Isn't hybrid multi-cloud exactly aligned with the future that you are portraying here (sans IBM's bet on k8s, which we could have another debate on) that what runs on top of cloud infrastructure will be available across multiple clouds?
Disclaimer: I work for IBM, and my opinions are my own.
Great! Reminds me: Is there any way each algorithm can keep a fixed colour/pattern across the graphs? For example: Annoy is grey on the first plot, but green in the second.
In either case, thanks to you Martin and Alec for the great work!
1. Diagonal moves have a slightly higher weight.
2. I have an edge weight of 100 to go through another person. This makes it possible to go _towards_ the right direction even though there is no path
3. For the perpendicular lines method, the distance is modified to 1e-3
4. A* doesn't work because the heuristic becomes pretty useless when you have moves that are very "cheap": you need a lower bound that's pretty much zero
So for all those reasons, I just went with plain old Dijkstra!
> A* doesn't work because the heuristic becomes pretty useless when you have moves that are very "cheap"
I don't see where you have cheap moves. If your move costs are unit for straight lines, slightly longer for diagonal (i.e. probably close to sqrt(2)), and much longer for obstacle, then A* will work fine. The lower bound is just the usual Cartesian or Manhattan (depending on your diagonal cost) distance.
The mechanism would work this way: sales people exhibit multiple features, and they are promoted based on some combination of those. If a sales person has outstanding other credentials, they might be promoted despite poor sales percentile. Those other credentials might actually be better predictors of managerial experience. Conversely, many of the top sales people might have been promoted on the grounds that they were good sales people, without exhibiting any other skills.
Note that there might still be a positive correlations between sales skills and managerial skills, but due to how the promotions are selected, you end up observing a negative correlation in the promoted group.
If Google knows that success in programming competitions is negatively correlated with job performance, then why Google organizes codejam and invites participants with good results to a job interview? Also as mentioned above Google interview questions looks like problems from programmings competitions.
True, and someone who is a great salesperson probably has skills that are optimized for the context of sales. But another aspect of getting promoted is that a person is given oversight of new kinds of activities, where teams work with different cultures and different rules. I've seen some sales manager succeed by being bullies to their sales team. But I think it is a disaster when someone attempts to bully a tech team. So what works in one context fails in another context. I've tried to describe this previously:
----------------------------
Every industry has certain euphemisms for the least savory aspects of its business. In sales, there is the secretly ugly phrase, “goal-oriented.” That sounds pleasant, doesn’t it? If I point at a woman and I say, “That entrepreneur is goal-oriented,” then you probably think I am complimenting her. But if I point at her and say, “That entrepreneur is a lying, manipulative, soulless psychopath who brutally exploits labor from the eleven-year-olds she employs in her sweatshops in Indonesia,” then you probably think I am insulting her, unless you are a libertarian. And yet both statements mean about the same thing: that she is someone who is willing to do whatever is necessary to ensure the success of her business.
When I read about Milburn online, I’d seen testimonials from his colleagues in which he was often described as a goal-oriented salesperson. That probably meant that he was a master of manipulating other people’s emotions. He knew all the tricks: praise, shame, laughter, anger, promises, guilt, threats.
Whether his use of these tools was conscious or unconscious is, of course, unknowable. But it doesn’t matter much. A lifetime as a sales professional left him with an arsenal of psychological ploys that had become second nature to him.
...Milburn truly had a genius for the strategic use of anger. If he sensed the risk of losing control of the conversation, he would indulge in another outburst. If I were to ever switch over to the Dark Side, I would want to study with him. His techniques were fundamentally dishonest and manipulative, but that is probably what made him so good at sales. And his tactics were probably an effective way to drive a sales team, but I sincerely believed that such tactics were the wrong way to run a software development team. Especially when doing something cutting-edge original, like we were doing, I think open and honest communications were extremely important. (I have worked with many companies where the sales team was both friendly and successful. One does not need to use abusive tactics to have success in sales. Indeed, the sales manager who relies on abuse is typically more interested in aggrandizing their own success, rather than the success of the company they work for.)
https://en.wikipedia.org/wiki/Around_the_World_in_Eighty_Day...
The following day Fogg apologises to Aouda for bringing her with him since he now has to live in poverty and cannot support her. Aouda confesses that she loves him and asks him to marry her. As Passepartout notifies a minister, he learns that he is mistaken in the date – it is not 22 December, but instead 21 December. Because the party had travelled eastward, their days were shortened by four minutes for every degree of longitude they crossed; thus, although they had experienced the same amount of time abroad as people had experienced in London, they had seen 80 sunrises and sunsets while London had seen only 79. Passepartout informs Fogg of his mistake and Fogg hurries to the Club just in time to meet his deadline and win the wager. Having spent almost £19,000 of his travel money during the journey, he divides the remainder between Passepartout and Fix and marries Aouda.