That's great! Take the opportunities you have to do good for the world. Everyone's circumstances will present them with different ones.
2,803 karma · joined July 22, 2012
Let me tell you a little bit about medical software. It is incredibly hard and expensive to get data. It usually involves finding enough people who are dying in a particular way (already hard), and managing to collect comparable data from each (i.e., with the same process: even harder). There is never enough of it to do the statistics you want. Once you have some, and you maybe come up with some way to do some step of the treatment process slightly better, there are a lot of hurdles and it takes a very long time before anything you did can actually be used on a patient. Rightfully so. People's lives are at stake.
With cancer in particular, maybe after a decade in the field you might know if what you did had a positive impact on treatment outcomes. Maybe. That's the best-case scenario. For me, I don't even know if any line of code I ever wrote was actually used to treat a real patient. Oh, yeah, and those dying people whose data you used to come up with and test your idea? Yeah, you could not do anything for them. They got whatever treatments were the best available at the time, and good luck. If they were great treatments, you probably would not have been working on the problem. So that sucks.
Meanwhile, at a browser company, I was able to write a patch and within a month or three it was deployed to over a hundred million people. Software projects I worked on are currently used to encode and decode a non-trivial fraction of the bits on the internet. They are deployed on every Android handset and every iPhone (not just in Firefox), so that's a few billion devices. Every time you make a Zoom call or use probably almost any other video conferencing service, you are using at least one, if not several, of those projects. I understand people are doing a lot of video conferencing these days. Maybe even some telemedicine calls between radiation oncologists and their patients.
So, uh, yeah. I feel like that had more impact. Not to discourage anyone from trying to cure cancer. It's incredibly hard.
P.S., "cancer" is actually a thousand different diseases with a thousand different causes and requires many different treatments. There will never be a "cure for cancer".
I have worked on software for treating cancer, and I have worked for Mozilla. And boy, let me tell you, the latter sure felt like it had a much bigger impact on the world.
The problem with patents and standards is that people are trying to charge for that value (the value of the network effects), and not the value of the technology. The big ideas of video compression are mostly more than 30 years old now, but the new patents are mostly on slight tweaks or additions (or things like header flags and other nonsense). No one would pay money for them except for the fact that they are part of the standard.
Source: I am a co-author on RFC 6716 and RFC 7845.
Yes, all of those things have been problems for a long time. However, the application of patent law to software and business methods is relatively recent (in legal terms). In the US, the Court of Appeals for the Federal Circuit has been largely responsible for growing the scope of software patents since its creation in 1982 (e.g., holding that use of an analog to digital converter or a memory storage device were all that is needed to meet the Supreme Court's 1981 Diamond v. Diehr test that a software patent has to actually do something, rather than just being a conventional implementation of inherently unpatentable math). This culminated in the 1998 State Street decision (again, of the CAFC, not the Supreme Court) that opened the floodgates for business method patents.
Despite Supreme Court decisions like In re Bilski (2009) and CLS Bank International v Alice Corp (2014) that attempted to limit the scope of these patents, and that were hoped (of Alice in particular) to be able to be used to invalidate large swaths of such patents, the CAFC continues to erode the force of these decisions (see, e.g., the 2020 Uniloc decision) and carve out new ways to patent mathematical algorithms implemented on general purpose computers. The allowance of these patents by the USPTO has also been on the rise in recent years.
Designing a development process that naturally achieves a malicious outcome is just as bad as actively making malicious decisions: https://www.zdnet.com/article/former-mozilla-exec-google-has...
This is what I would predict would happen with any UBI, more generally: it would just raise the prices of the things that people have to buy until all the money wound up with those who had the capital and the sophistication to capture it. It's not clear you'd even wind up with more of those things you listed, but it is certain you'd pay more for what you did get.
"A single prediction that you can take the difference with" makes some assumptions about the shape of your distribution, at least under any reasonable model of "code the difference" where, e.g., probability decreases as the difference gets larger. These are often very good assumptions, to be fair.
Of course, your DNS is likely still unencrypted. But maybe one day we'll get there.
I agree that part is not really unique, but if you are an obvious foreigner and stare at a map in a train station for more than a few seconds, a stranger will come up and ask you if you need directions. Maybe it is just people who want a chance to practice their English, or something (since they all have to learn it in school, but do not get a lot of chances to use it in daily life). However, I have literally had someone walk me halfway across the station and wait with me on the platform to personally see me get on the right train, during the middle of commuter rush hour, even though it was not the way she was going at all. That kind of stuff only happens to me in Japan, and it happens to me there repeatedly.
Of course, in the modern cellphone era, navigating Japan is not the imposing challenge for foreigners that it once was.
In my experience, it is not as often as you might think. The key issue comes down to brilliance coupled with selfishness.
If you look for it, you can see some common themes in these comments: "...the other team members were all forced to work around his refactor and waste time updating their PRs to match his work," or "Meanwhile we were all left to deal with his fallout," or "...this particular eng ... ended up costing almost as much to clean up his mess as the value we got from him in the first place." We have all had to work with that guy. In the early part of my career, I have been that guy.
People are not arguing because they don't understand your brilliance and are jealous of your productivity. It is because you are making their lives harder. That is what makes people bitter.
Turn the problem around: instead of asking "how can I get all of these over-zealous, jealous people to leave me alone", ask, "how can I apply my brilliance to make their lives easier instead of harder?" Unless you are doing something basically trivial, there is usually too much work for any one person to do it all by themselves, no matter how brilliant. You need those people. So help them out. You will have a much better time. And so will they.
For a fixed budget, assuming that Moore's Law still holds (dubious, these days), you should wait until the computation will take 26 months on hardware purchased within that budget [0].
And maybe you're fine with that today. But once you've ceded that ground, what happens when they move the line?
Because they always move the line.
It actually does change the dates for when patent documents are considered prior art (as described in the first paragraph I referenced), so it is easy to get the issues confused.
The effects of first to file on how prior art works are, in my opinion, way more important than the litigation edge condition it eliminated.
In particular, the paragraph starting with "Although some of the provisions of AIA 35 U.S.C. 102(a) and (b) are similar to pre-AIA 35 U.S.C. 102(a), (b), and (e), the AIA has introduced a number of important changes with respect to prior art ..."
Edit: And more importantly, Section 2152.01 right below it.
In either case, it's likely that the USPTO isn't going to notice before issuing the patent, but under first-to-file it is way easier (read: cheaper) to show that "X document was published on Y website on Z date" than it is to start doing discovery to compare private lab notebooks or whatever to decide who really invented the thing first.
This only ever affected a minor number of cases anyway, and the change mostly just made litigation costs cheaper and more predictable.
It also seems really great for open source software (where I personally work), since everyone publishes everything they do immediately anyway.
As someone who did their Ph.D thesis on the statistics of shape using models based on the medial axis (i.e., a skeleton), I would beg to differ.
Whether these models are as easy to apply (computationally and conceptually) as the currently in-vogue techniques is another question, but there is nothing magical here that computers are incapable of.
Intra paint was not defeated because of the cost of entropy coding "don't use feature X here", it was defeated because due to "lapping and block size alignment issues, we cannot just turn it on and off when we want" (last paragraph of the "Pretty Images" section).
It is also not just beating AVC. They claim that it is higher quality than their own HEVC work, which they invested in heavily before it become clear that the licensing situation was not going to just "work itself out". They once told us that even they were surprised how many of the HEVC tools you could wind up using for real time, if you used them judiciously (basically: all of them).
I have worked with some of the team responsible for this project before, and we collaborated with them on some AV1 tools (most notably the CDEF loop filter, but also entropy coding and a few other things). They know their stuff, and really do seem to be leaps and bounds ahead of everyone else on the RTC side of things.
I would agree, but not for the reason you state. IE was the standard back then, too. The difference is that once it was "good enough", Microsoft reduced investment in it and allowed it to stagnate. Others caught up in terms of compatibility and passed it in terms of usability.
Google, on the other hand, invests so much that no on else can keep up, and knows that this is the means by which they maintain control over the web (and web standards).