If you expected to learn useful finance at a university, you already failed as soon as you applied. If anyone there knew what they were doing financially, they wouldnt be there, and especially not teaching Finance 102. In case you were not aware, nearly all professors hate teaching lower level classes, so they naturally get forced on the worst performing professors (the ones that are closest to getting fired).
Concluding that the entirety of TA can not be useful (the Null hypothesis is true for all input signals) just from seeing that a few strawman TA in finance 102 not work is beyond absurd.
Absolutely but there are cases where doing that is not practical. For instance machine learning training tasks that can take a long time to converge (months even). There are good arguments that long running tasks should not be in a notebook but they often find themselves in notebooks regardless especially in the ML domain.
Any product/service directly and publicly marketed to criminals is a honeypot (or will soon be). There is no reason for a real business (legal or illegal) to prefer criminal customers in good faith as all they bring is problems.
There are many agencies around the globe that set these up and continue to operate after siezing them to catch the low hanging fruit criminals. Its a lot easier/cheaper to lure criminals expose themselves directly than it is to actually investigate/track/hunt them down. Its pretty much the quantity over quality approach.
Its potentially an even worse problem than overfitting because of the error accumulation and undermining of the fitness functions. Its similar to asking students to create test questions to evaluate themself. As the proportion of self output questions grows, the context eventually drifts to be meaningless. In other words it leads to test "questions" like "The answer is A".
One measurement of this would be to watch the growth of probability of new models to say similar things to older model filters like "As a large language model ..."
Have sold to a lot of governments and large contractors. I wouldn't say its a death sentence, just different business model than most startups. Government and large orgs that send out RFPs are essentially asking for a product/service to be made and maintained specifically for them. Unless your essentially in the consulting business, its rarely worthwhile for a normal startup as it eats A LOT of time for a single customer. Government can be very lucrative if you can find the right niche but its always exceedingly painful and one of the governments favorite pastimes is completely 180ing requirements right after you complete them.
The Federal Reserve: Don't worry my friend, its all just transitory due to <insert latest event>. *Slaps side of money printer* These babies can print us out of any hole!
Nice try FBI. In all seriousness though, has it actually gotten so bad that yandex of all search engines is less censored? Or is it just less censored when it comes to topics controversial to the US (and not russia)? The fact that so much censoring is going on that google has a "hold on while we censor this" page is insane.
It may also be part of their anticompetitive war on other browsers. I get captchas constantly in a new default Firefox profile, but not in a new default chrome profile. Spoofing user agent to recent chrome agent in Firefox makes the captchas happen far less often for me.
I look forward to the day the manager can have a daily standup/progress report meeting oblivious to the fact that (s)he is the only one actually present... One step closer!
All of those happen on VPNs period, not just with port forwarding.
Dealing with annoyed law enforcement, hosting providers, and IP reputation is 99% of the value of a VPN. The other 1% is just setting up a VPN server to open proxy everything (which there are scripts on github that can do it in 2mins). Of course its not really preserving privacy much unless there are multiple users...
Any significantly shared connection will have at least one person abusing it and causing most of the problems, the logical conclusion would be to ban the few abusers but if mullvad truely doesn't log/retain billing data as they claim, permanent banning would be difficult as a new account could just be created.
I don't see why they couldn't do some kind of compromise like an account has to be of certain age/spend to use port forwarding. They do keep mappings of ports to account, so its not like they don't know which accounts are abusing. Getting banned would then be more expensive for the abusers.
I like to use rsync.net for backups. You can use something like borg, rsync, or just sftp/sshfs mount. Its not as cheap as something like S3 deep (in terms of storage) but it is pretty convient. The owner is a absolute machine and frequently visits HN too.
It may depend on the makeup of data or something. They "requested" one of my prior projects go on the enterprise plan after about 50TB, granted the overwhelming majority of transfer was for distributing binary executables so I was in pretty blatant violation of their policy. This was 2015ish, so the limit could also have gone up over time as bandwidth gets cheaper too.
Unless its 100TB/mo of pure HTML/CSS/JS (lol) cloudflare will demand you be on enterprise plan long before 100TB/mo. The fine print makes it near useless for any significant volume.
I agree with you about his descriptions, they certainly give off the exaggeration feel. I am just saying that I also wouldn't be surprised if someone actually tried to swat him (among other things). I've seen him several times start a fight for practically no reason and continuously provoke it further, swatting really isn't that far down the line of provocation when it comes to the internet sadly.
I'm not aware of what others have done to him but his claims honestly wouldn't surprise me if they were accurate as he is quite well known for being publicly very rude regarding GrapheneOS.
My first interaction with him was asking about how to backup all data on GrapheneOS (including secure/restricted data). He flat out called me an idiot, said that it "defeats the point of grapheneos", and refused to help. Most of the help formats that he is available on are full of responses very similar to this from him. Hardly by any stretch of the imagination professional conduct.
I do love GrapheneOS and I'm sure managing it as he does/did is quite stressful. I don't have any Ill will towards him, just concern for him and the project. I think stepping down and getting away from social media will be a great move for both him and the project. He clearly wants some much needed time to recover and destress. Frankly, his public behavior was the biggest liability for the project but him stepping down is a good sign that at the very least he is making moves to start recovering.
Just FYI google and bing publish their user agent strings[1][2] for the crawlers. At least in my experience most of the typical ad-infested and paywalled news sites wont display the paywall if you change the user agent to a crawler they prefer.
While I would agree the vast majority of the time, ads are also used heavily by small businesses and independent developers to gauge interest and even just to keep the lights on sometimes. Mostly referring to online ads, billboards and TV ads are much more expensive and are far more risky for small businesses.
Not knocking Ki or any new language but what does it offer over existing systems? I see the criticism of a few languages on the website but they all boil down in reasoning to things like: garbage collection = bad, makefiles = bad. It also doesn't mention or demonstrate how Ki addresses the complaints. For example: if Ki doesn't do garbage collection or manual memory management, what does it do? Smart pointers/reference counting? How are circular references handled? (E.g. doubly linked list)
Would recommend removing the "Other" comparison as pretty much anything that can be said there is wildly inaccurate. Would also like to see comparison with Zig as it seems to have similar goals.
Concluding that the entirety of TA can not be useful (the Null hypothesis is true for all input signals) just from seeing that a few strawman TA in finance 102 not work is beyond absurd.