Posts by Christian Berger

To understand this, one must take this in the context of "long termism"

"Long Termism" is a sort of "silicon valley religion".

The basic idea is like this:

People believe that eventually there will be an "all powerful superhuman AI" which will save us all, perhaps by resurrecting us all as some sort of simulation in a digital heaven. Perhaps spread over multiple solar systems, etc. This will provide heaven for hundreds of billions of people.

If you don't help advancing the progress of such a system even though you could, that system, which will eventually happen, will punish you retroactively for this... for example by creating a digitally simulated copy it'll torture sometime in the future.

Suffering of billions of people in the present is completely acceptable in this world-view as it will save hundreds of billions of people in the future.

Of course this is all just a form of religion, also Altman probably will delay anything resembling "intelligence" by some decades. There is no reason for such a system to be spiteful, etc.

It's kinda like world religions, if you take all the positive ideas and values away. You have no "love your neighbor", but are still left with money hungry pedophile billionaires.

Isn't it wonderful?

Yet another business model poking the bubble, scaling down the AI business to something more sustainable.

I mean computer programming is certainly not something those text generators are particularly useful at... particularly not at those prices. $200 is a lot of money when compared to programmer wages.

Well the problem is unneeded dependencies

Solutions like PyPi make it seem like having dependencies is easy and no problem, by hiding the problems. This makes people use more and more dependencies, making the problem much worse.

That's why I try to avoid them. If I write software in Python and a library is not available in Debian/Ubuntu, I will not use it.

Well modern SaaS doesn't care about about quality

Once you've got your customer, quality no longer matters. A prime example is Jira, a software so bad, screens need _seconds_ to load, are confusing... and sometimes you loose all of your work because of a wrong button you've touched.

This is a logical cause of the SaaS concept. You have subscriptions which means there is no motivation to keep the user to want to pay for an upgrade. You no longer need to make your software better in the next version as people are already giving you money, regardless of quality. If you are just entrenched deeply enough into a company, quality is completely irrelevant, as people will not switch, even if they want to.

It seems to be on brand

I mean... we all suffer from client side React.. which somehow manages to turn trivial GUIs into something that takes _gigabytes_ of RAM and multiple CPU cores of processing powers... even when idle....

It only seems on brand to also mess up the server side of things in spectacular ways.

Well I partially blame Big-"Tech"

In order to find simple solutions, you need to think about a problem before you actually start coding. Instead, if you are Big-"Tech" you can just throw dozens of developers onto any trivial problem.

The big advantage, from the point of Big-"Tech" is that you can increase the complexity of standards, raising the barrier of entry for competitors. That's why things like HTTP/2 and HTTP/3 exist.

Factual error in the article

The 80186 and 80188 chips were actually not really in between the 8086 and the 80286, but instead more microcontroller like devices. While they didn't have RAM or ROM inside, they had some peripheral devices like a timer and interrupt controller on board. Unfortunately those were incompatible with the IBM-PC so they were essentially useless for the PC market. Some manufacturers like Siemens used them for their DOS-compatible PC-D lines.

However since cross compiling wasn't to hard, the 8018x processors were heavily used in the embedded sector, for example for ISDN PBXes.

So since the proof of personhood is via an iris scan...

... it's probably fairly easy to build a device that can just fake lots of irises, gaining $16 a pop. Plus it's ethically acceptable since you are only defrauding crypto scamming billionaires.

I recently demoed Lazarus at our company

Essentially while explaining Lazarus I was able to click together a basic CURD application within 8 minutes... without touching the keyboard.

That's like going bungie jumping...

... and then suing the bungie jumping company for tying an elastic rope on you and throwing you of a bridge.

Everybody knew that such software is highly problematic. It's simply not a good idea to try to fix the problem of to much software by adding more software... particularly if that software is written by people who are not security minded.

I can understand private people being fooled by companies like this, but we are talking about a large company... with legal departments. Why didn't the legal department find the clauses that said that the software must not be used for critical applications? Why didn't any of the technical departments object to that sort of software? Why wasn't anything done when the Linux version of that software had, essentially, the same bug... some weeks before this?

Well 16 and 32 bit support may important

I mean, sure as someone who runs Linux or *BSD on a server, it doesn't really matter if you have 16 and 32 bit support... but then again, it doesn't really matter if you have x86 or ARM or whatever.

The area where this doesn't matter is the Windows-Server market. There software is distributed without source code... and backwards compatibility is vital. The portion of that market that can't just switch to normal operating systems is "application servers for legacy applications". That old business-critical 16 bit WinAPI Application the company bought in 1993 will certainly not run on ARM.

I mean it's "Secure Boot"

It's a highly risky technology, originally designed to enable censorship for operating systems which, at best, protects you from some highly selected attacks.

I wonder if it's possible to create barcodes that way

I mean it would be a cheap way to transmit information to mobile phones by using an LED and a diffusor. You could create barcodes that could carry information like an authentication code or something.

Strictly speaking it's their Z84 line, not the Z80 line

The Z84 are CMOS variants which are faster and use less power than the original Z80 chips while being pin-compatible. The original Z80 chips probably have been discontinued decades ago.

That's why you still get Z80-flavoured systems based on even newer implementations. The Z84 chips kinda are overpriced with the CPU alone selling for more than a whole 32-bit micro controller fast enough to emulate it.

Well it's a sign of the crapularity

We do essentially the same things now as we did in the past, but with a lot more code. More code means more bugs resulting on the every day experience that computers don't work very well.

There is some hope that this comes in waves, like the terrible software we had in the 1990s being washed away by obsolesce and good ideas like the UNIX philosophy becoming more popular again. It may we that we are just in a current wave of that.

However if we aren't and the current trend continues, we might be heading towards a "Crapularity", a singularity of crap, where the systems around us keep failing and nobody can mend the mending apparatus as in E.M. Forster's "The Machine Stops"

There is one important legacy from it...

and that's UTF-8 which strips away multiple layers of complexity when dealing with international texts. There are now a lot of cases where you can just handle text as text, not having to worry about whether the characters inside of it are hieroglyphics or Arabic characters.

It's an attempt at turning back the time

Instead of trying to find innovative ways to get money to creators, the DMCA just tries to turn back the time to the 1970s, claiming that it can keep people from using computers. Essentially they try to maintain business-models that make no sense in the computer age.

In some areas, like mobile phones, they even succeeded, making manufacturers lock down their systems so users no longer can easily upgrade to another operating system. That is why there is so little progress in that field.

I once was able to test them

... they are almost comically bad.

I mean one of the main tasks of such devices is to capture 20ms of audio samples, package them into a RTP frame and then send them. There's more to building a good ATA, but that's the most basic thing you need to do.

Of course, since they are running normal operating systems, there is a bit of jitter to be expected, so if you compare the expected time of arrival to the actual time of arrival you should see some noise, in a histogram that should resemble a normal distribution close to zero. So the error is more likely to be between 0.1 and 0.2 ms than between 10.1 and 10.2 ms. It may be overlaid by a steady rise or fall because your sampling rate is not precise, but any half decent ATA will compensate for that.

The adapter I've been able to test had a "box shaped" distribution... and to make matters worse, the culminated error rose in discrete steps. Each one of those steps means that not only you will have a glitch in your audio, but also very likely that your modem connection will break. I have no idea how on earth they managed to make their device that bad.

It actually doesn't matter which megaorganisation wants to controll the Internet

I mean we also see "NEW IP" movements from the west, with Big "Tech" wanting to take over essential services like DNS or the Web.

It doesn't matter if those organizations are motivated by greed or the desire for control, we need to oppose it.

The obvious way to do so is to avoid complexity. Simple protocols are easier to implement. That means you have more implementations and less control from big organizations.

Plain IPv6 is a good step in that direction as it simplifies a lot of things. For example there no longer is a need for NA(P)T which eliminated the need for workarounds for it. You can have end-to-end connectivity which eliminates the need for central servers. You could build things like messaging systems without any need of an organization operating it.

Yeah, but by NFT logic...

... I can prove that nobody was harmed when the Titanic sank.

There are immutable pieces of paper that were distributed thousands of times that claim "TITANIC SIKNING ; NO LIVES LOST"

https://www.huffpost.com/archive/ca/entry/titanic-headlines-in-vancouver-got-it-very-wrong-photos_n_4178419

Then there is this Italian movie, distributed in immutable DVDs to thousands of people, The Legend of the Titanic (1999). This movie claims that nobody died, as the octopus that was responsible for the ice berg being in the way, saved them all.

It's not like any "Blockchain" is more authoritative than a random newspaper or a movie, except in the eyes of some idiots.

I'm not sure if they have ever seen a rocket launch

Considering that you need huge rockets filled with propellant that will be turned into CO2 and water in minutes just to send up a shipping container full of equipment, I doubt that this will ever be able to offset the environmental cost of producing electricity on earth.

Also on earth you can just use solar power and wind to power your datacentre.

Why would anybody accept this?

...except for maybe those people who are to bad at what they do to not find a new job quickly?

I mean, Twitter isn't infrastructure. If it vanishes tomorrow few people will even notice.

It actually could be much less

Such services typically determine market shares by trying to run 3rd party Javascript in browsers. Since running extensions like noscript is correlated with running Linux or *BSD, those operating systems are likely under represented in those studies.

There is a good point for this, heise.de, a website mostly browsed by people at work (=>much higher Windows percentage than global) still has lower numbers for Windows than those statistics.

How does one secure it?

I mean, sure it might be meltdown-proof, but what if someone goes there and either just takes the reactor with them or plants a bomb next to it.

Considering there are decent amounts of radioactive materials inside of it, using it as a "dirty bomb" seems a plausible scenario.

Simply put: There aren't many jobs in that area

For example Germany has jettisoned most of its non-automotive companies in the last few decades. There are no solar companies or communications companies any more. Those that do exit have outsourced most of their technical aspects. Most of what's left is in the automotive industry, but you don't want to go there.

To be honest my decision to study electrical engineering was mostly influenced by it's use as a hobby.

Well we knew that their use was rather limited

However, at least in Germany they actually have turned into a beacon of hope, but not in the way one would expect.

The German tracing app CWA (Corona Warn App), not only followed the minimal standards you would expect, like having public source code for the app, no they have public source code for all components including public documentation on the rationale behind the scheme they used.

And the design was very decent trying to keep the advantages without compromising privacy. Essentially it was done in a way you only had to expose yourself when you got tested positive.

Now the main point of criticism was that it was hugely expensive... until the Luca App came along which claimed to do the same thing, but was developed a lot less competently yet charged a similar amount. While the CWA people actively engaged with the security community, for example by giving talks at hacker conferences, the Luca people only complained that people were complaining about basic defects of it.

Pasives are not semiconductors

Particularly not resistors, inductors and capacitors.

Passives are more or less defined by not being semiconductors. (there are of course exceptions and gray areas)

Nobody expects that kind of software to do anything other than...

...waste CPU cycles. After all the only remotely sensible feature (scanning for known bad strings of octets) is only semi-useful so vendors try to proof maths wrong by trying to find out if code is "good" or "bad" by just looking at it.

Well why bother with it at all?

I mean after all those are devices where actual security is more important than any kind of security snakeoil. Just build them in a sturdy and locked case and have decent hardware interlocks to prevent bad things from happening.

If bad things can happen from errors in the software, secure boot will not help you much, as in order to change the boot process, you already need very high privileges. On the other hand, the charge process, which can cause harm if interfered with, has to have interfaces so the outside world.

Well the requirements of loading data overlap with the requirements for audio

Essentially audio tape formats in the home computer era work by storing pulses on the tape. This was done in order to keep the hardware simple and to be able to deal with extremely horrible tape decks running at the wrong speed.

If you manage to keep a pulse vaguely coherent, you will be able to get the data through. This is also called a linear phase response, or a constant group delay. This isn't as important for general audio, so back in the analogue days people didn't take the effort of doing so.

Now since everything is digital, it's easy to have filters with linear phase response.

Also any general purpose audio codec has very little reason to mess with the phase, so the data will likely go through given sufficiently high bit rates.

BTW back in the analogue days there was a variety of ways to deal with non-linear phase responses which you could have when the audio was transmitted via television or the radio. The WDR Computer Club (West Germany) had a device which would cup off the overtones of the tape signal. That way they didn't need to care as much about what the overtones contributed to the signal. In (communist) East Germany, the solution was to broadcast the signal in stereo, loud on one channel and with 10dB attenuation on the other one.

Some time ago I listened to a talk from one of the Kubernetes proponents

It was a talk about security issues, and when it came to the issue of over boarding complexity often caused by Kubernetes he said that there is a difference between "what people want" and "what people need".

Seriously if you want complexity maybe you shouldn't work in IT, and area where complexity is the main problem.