The Ideal Internet: Point of View Is Important
September 11, 2020
I read “Now the Impact of Regulation on the Internet Can Be Gauged.” Interesting but fanciful, the article lays out what the Internet should be. The main points appear to exist in a mental construct removed from political turmoil, the Rona, and financial challenges.
The write up explains that the Internet Society has crafted an Internet Impact Assessment Toolkit. I learned that the:
Internet Way of Networking (IWN): Defining the Critical Properties of the Internet, … explains how the Internet’s unique foundation is responsible for its strength and success. It also identifies the critical properties that must be protected to enable the Internet to reach its full potential….The Internet Impact Assessment Toolkit is a guide to help ensure regulation, technology trends and decisions don’t harm the infrastructure of the Internet.
Here are the key elements of the IWN:
- An accessible infrastructure with a common protocol – A ‘common language’ enabling global connectivity and unrestricted access to the Internet.
- An open architecture of interoperable and reusable building blocks – Open infrastructure with a set of standards enabling permission-free innovation.
- Decentralized management and a single distributed routing system – Distributed routing enabling local networks to grow, while maintaining worldwide connectivity.
- Common global identifiers – A single common identifier allowing computers and devices around the world to communicate with each other.
- A technology-neutral, general-purpose network – A simple and adaptable dynamic environment cultivating infinite opportunities for innovation.
Quite idealistic, and the statements do not address the reality of corrosive social networks and the emergence of corporate nation states. And there’s China. Oh, right, China.
Stephen E Arnold, September 11, 2020
Yo, Kafka: Check Out This Bureaucratic Play
September 9, 2020
“Beijing Floats a Plan to Protect Chinese Companies from American Cyber Bullying” is an interesting news report. Let’s assume that it is accurate with nothing lost in translation. The write up states:
In a speech Tuesday, Chinese State Councillor Wang Yi proposed a set of international rules intended to increase trust and refute the Trump administration’s strategy to limit the reach of Chinese-made technologies. Wang said the “Global Initiative on Data Security” is a recognition that data protection techniques are increasingly politicized at a moment when “individual countries” are “bullying” others, sometimes “hunting” foreign-based companies.
The political questions are outside the scope of DarkCyber. The semantic issues are getting into the research team’s area of interest.
What’s important is that this is a content object which may be weaponized. Who is bullying whom? Has security become the equivalent of accosting a person of improper behavior? What’s hunting mean?
Worth noting.
Stephen E Arnold, September 9, 2020
Names: A Problem beyond Math
September 7, 2020
I read “Why Mathematicians Should Stop Naming Things After Each Other.” The main point is that naming conventions in mathematics make it difficult to know to what something refers. There are numerous examples; for example:
A Calabi-Yau manifold is a compact, complex Kähler manifold with a trivial first Chern class.
A possible explanation? Consider:
The memory-intensive naming schemes in modern math may have the result of boxing out the laymen, but we must hope the priests of the academy are not doing it on purpose.
My view is that making names tough to parse adds some magic and special sauce to what might otherwise be a “so what?” insight. On the other hand, weird naming prevents meaningful connections to be perceived. What if a Hopf fibration is related to giant waves in the universe? Making the connection is tough with today’s naming policies.
My personal view is that many experts are nervous about the validity and value of their research or insights. Hiding behind language and naming conventions deflects criticism.
The same approach fuels the use of jargon and techno-babble. Search is not find. Search is discovery. Yeah, right.
Stephen E Arnold, September 7, 2020
Backing up Your Wonderful Mac: Think Twice
September 6, 2020
If you have a Mac anything, check out “Under Construction.” The write up does a good and mostly politically correct explanation of why one’s Mac back ups often disappoint. People love Apple and adore their Mac whatevers. We were testing a software which bonds two or more Internet connections. The software was okay, but the zippy super duper antenna we had to purchase was a bit more problematic. To make the painful two days short, the drivers to make the wonderful Mac connect to the external super duper antenna did not work. The installation nuked the existing installation of our former friend Catalina. We figured no big deal. We had TimeMachine. We had a manual back up on a separate external device.
Problem? You bet.
Nothing restored. If the information in Under Construction is on the beam, the problem is a result of Apple’s lack of engineering attention.
What was our fix? We reinstalled Mojave and got everything working. The TimeMachine thing? We are not using it any more. We manually copy data files and will rely on complete reinstalls when the wonderfully mis-engineered Macs create such excitement. Back ups? Hey, no big deal. Buy an iPhone.
Stephen E Arnold, September 6, 2020
Intel: Why the Horse Collar Outfit Stumble Is Wobbling
September 4, 2020
A story in Techradar helps illuminate the magnitude of the Intel wobble. Right, Intel. The outfit which has watched a smaller firm zip along in the desktop CPU market and repeatedly miss chip deadlines. Even Apple has reduced its dependence on the iconic company which uses buzzwords instead of silicon and more with it materials to make its point. “Here’s How Huawei and Other Chinese Firms Could Access Crucial CPU Technology without Restrictions” contains an interesting paragraph which may be made up but may not be:
The number of Chinese chip designers skyrocketed from 736 in 2015 to 1,780 in 2017. Many of these companies need CPU IP and some may not be inclined to use Arm. For them, MIPS and RISC-V architectures are two natural choices and MIPS has an edge over RISC-V right now. MIPS does have off-the-shelf high-performance CPU cores comparable to Arm’s Cortex-A70-series or Neoverse, but it is possible to use the MIPS architecture to build something powerful enough for servers. For example, China’s Loongson Technology develops MIPS64 CPUs for client devices and servers and there are also Green500 supercomputers based on MIPS CPUs.
Intel seems to be managing in a way that inadvertently makes its competitive posture scleotic.
Stephen E Arnold, September 4, 2020
Intel Code Names: Horse Feathers, Horse Collars, and Fancy Dancing
September 3, 2020
Intel loves code names. And what a knack for coinages? Pentium. What’s not to like. I noted this item last year (2019) I believe:
Intel, in partnership with the U.S. Department of Energy, Argonne National Laboratory, and Cray, is building the nations first Exascale supercomputer. By accelerating the convergence of high performance computing and artificial intelligence, Exascale supercomputing will advance scientific research and enable breakthroughs in neuroscience and cancer research, aerospace modeling and simulation, and theoretical research of our universe. The Aurora system will be based on the future generation of the Intel® Scalable Processor, the future Intel® Xeon® compute architecture, the next generation Intel® Optane™ DC persistent memory, and supported by Intel’s One API software.
Note the word choice: Convergence, high performance, artificial intelligence, Exascale, super computing, modeling, simulation, theoretical research, scalable, Optane, and One API.
Do I have a problem with this English major with a minor in marketing writing? Nah. Makes zero difference to me. We switched to Ryzen 3950X silicon. Workin’ just fine.
However, the venerable New York Times published “Intel Slips, and a High Performance Supercomputer Is Delayed.” That write up stated:
Intel, the last big US company that both designs and makes microprocessors, signaled in July that it might for the first time use foundries owned by other companies to make some cutting edge chips.
Now it’s September, and how is Intel doing?
Not too well. The Argonne Aurora supercomputer is delayed. Chinese computer scientists rejoice.
Is this Intel stumble important?
Yes, buzzwords and MBA speak cannot disguise the fact that Intel cannot deliver on time and on target. But, wow, Intel can spin fancy phrases; for example, Optane as in “Argonne can Optane its supercomputer.”
Another Covid moment?
Stephen E Arnold, September 3, 2020
Technical Debt: Nope, It Exists and That Debt Means Operational Poverty, Then Death
August 28, 2020
“Technical Debt Doesn’t Exist” is an interesting view of software. The problem is that “technology” is not just software. The weird behavior of an Adobe application like Framemaker can be traced to the program’s Unix roots. But why, one asks, is it so darned difficult to manage colors in a program intended to print documents with some parts in color? What about the mysterious behavior of Windows 10 when a legal installation collects $0.99 cents for an HEVC codec only to report that the codec cannot be installed? What about the enterprise application from OpenText cannot display a document recently displayed to the user of the content management system? Are these problems due to careless programming?
According to the article:
There is no such thing as technical debt. There is work to do, that we can agree on, but it’s not debt payment.
The punch line for the write up is that technical debt is just maintenance.
Let’s think about this.
The constraints of Framemaker result from its Unix roots. Now decades later, those roots still exist. Like the original i2 Analyst’s Notebook (a policeware system), some functions were constrained by the lovely interaction of the hardware, the operating system, and the code. The Unix touches remain today: Enter Escape O P C and the list of styles pop up. Yep, commands from 40 years ago are still working and remain inscrutable to anyone trying to learn the program. Why aren’t there changes? Adobe tried and ended up with InDesign. I would suggest that the cost of “fixing up” Framemaker were too high if Adobe could corral engineers who could do the job. Framemaker, therefore, is still around, but it is an orphan and a problematic one at that.
What about Microsoft and a codec? The fact that Microsoft makes a free version available for a person willing to put in the time to locate the HEVC download is one thing. Charging $0.99 for a codec which cannot be installed is another. Figuring out the unknown and unanticipated interactions among video hardware, software in the Windows 10 fun house, and third-party software is too expensive. What’s the fix? Ignore the problem. Put out some marketing baloney and tell the human doing customer support to advise the person with the failed codec to reinstall Windows. Yeah, right. A problem exists that will be around for exactly as long as there is Windows 10.
What about the OpenText content management system? We encountered this problem when trying to figure out why users of the system could not locate a file which had been saved the previous day. We poked around the hardware; we poked around the content management system; we poked around the search system which turned out to be an Autonomy stub. Yep, Autonomy search was “in” the OpenText system. The issue was the interaction of the Autonomy search system first crafted in the late 1990s, the content management system which OpenText bought from a vendor, and the hardware used to run the system. Did OpenText care? Nope, not at all. Open a file and wait 15 minutes. And what about the missing file? Updates sat in a queue and usually took place a couple of days after the Save command was issued. The fix? Ho ho ho.
Let me be clear: When a system is coded and it sort of works, that system is deployed. If a problem surfaces quickly, the vendor will have someone fix it. If it is a big problem, maybe two or three people will work on the issue. Whatever must be done to get the phone to stop ringing, the email to stop arriving, and angry customers to stop having their lawyers write nasty grams will be done. Then it is over. No one will go back and figure out what went wrong, make fixes, and dutifully put the ship in proper shape. The mistake is embedded in digital amber and the “fix” is part of the woodwork. How often do you look at the plumbing connections from the outside water line to your hot water heater. What happens when there’s a leak? A fix is made and then forget it.
What about technical debt? The behaviors I have described mean that systems persist through time. The systems are not refactored or “fixed”. The systems are just patched. Amazon enshrines this process in its two pizza teams. And how about the documentation for the fixes made on Saturday morning at 3 am? Ho ho ho.
Let me offer some observations:
- Significant changes to software today are mostly cosmetic, what I call wrappers. The problems remain but their pointy parts are blunted.
- The cost of making fundamental changes are beyond the reach of even the largest and most resource rich organizations.
- The humans required to figure out where the problem is and make structural changes are almost impossible for most technologies.
The article calls this maintenance. I think that’s an okay word, but the reality is that today’s software, particular software based on recycled libraries, existing systems accessed via application programming interfaces, and hardware with components with checkered or unknown pasts are not going to be “fixed.”
We live in an era of “good enough.”
The technical debt is going to catch up to those who sell and develop products. Users are already paying the price.
What happens if one pushes technical debt into tomorrow or next week?
That’s an easy question to answer. The vaunted “user experience” becomes more like a carnival act while the behind the scenes activity is less and less savory. How about those mandatory updates which delete photos, kill a Mac desktop, or allow a mobile phone to go dead because of a bug? The new normal.
It’s just maintenance. We know how much bean counters like to allocate cash for maintenance. Operational poverty, then the death of innovation.
Stephen E Arnold, August 28, 2020
About Process IBM and Intel Chips: Lame and Lamer?
August 28, 2020
AnandTech published “TSMC Details 3nm Process Technology: Full Node Scaling for 2H22 Volume Production.” Most people don’t know a nanometer from a Gen X tweeter. No crazy physics required for this post. What’s important are these two “big” announcements from US technology companies who are in the CPU business.
The first announcement is from Intel. That’s the outfit with the Horse Collar quantum computing thing. No, you can’t get one yet, maybe ever. Who really knows? Intel is now going to ship CPUs using 10 nm process technology with modern with it process technology scheduled for 2021. Let’s go with 7nm. I like assuming that Intel will catch up with AMD Ryzen 3000s. For “color”, you may enjoy this NYT write up about the Intel Inside crowd. Prepare to pay for “all the news,” of course.
The second announcement is from Big Blue. That’s the outfit with IBM Watson which also sells mainframes. (Thank goodness for the RedHat acquisition.) You can now purchase the really popular Power9 CPUs fabbed at 14nm.
So what?
If TSMC does move to 3nm in 2022, will IBM and Intel have a horse in the race? Moving the wonderful Intel architecture to parity with AMD has been — how shall I phrase it — a long, painful journey in a Yugo.
IBM has to move from 14nm to 3nm. Hey, just ask Watson how to pull this off.
With ARM, Amazon, and Chinese CPU outfits pushing in new directions, perhaps one should consult the oracle at Delphi about the future business opportunities for IBM and Intel. Pigeons work. Moving to more modern, energy efficient, and sometimes speedier CPUs may be a challenge. Where did that pigeon go? Taiwan and South Korea where the fabs are?
Stephen E Arnold, August 28, 2020
OnionFruit Revamps With New Browser Version
August 26, 2020
Remaining anonymous is impossible online, especially with all the cookies we “eat.” Instead of an all cookie diet, try using a browser made from onions and fruit! Major Geeks revealed their latest harvest with an update to their popular TOR browser: OnionFruit Connect 2020.730.0.
TOR browsers work, because they encrypt a user’s browsing data in many security layers like an onion. In order to identify the user, one has to peel back layers of encrypted data. It makes hacking someone with a Tor browser tedious and extremely difficult. TOR browsers also allow people to connect to the Dark Web that uses encrypted and random web addresses.
OnionFruit guarantees its users are protected:
“Having the ability to use a browser that you are already comfortable with makes using TOR more of a seamless process. OnionFruit Connect will initiate the TOR service and then configures your proxy settings allowing your apps to be routed through TOR’s tunnel. You will be notified that you’re protected, confirming that all your internet traffic is being passed through the TOR tunnel safely encrypted. This process ensures that every single site you visit gets routed through multiple servers to help mask your actions, making them difficult to track.”
OnionFruit is simple to set up on a computer and then access the TOR network. The best thing is that it works with favored browsers: Chrome, Firefox, Edge, Opera, and others without an extra configuration. OnionFruit updates itself, has custom landing pages, and a download speed monitor.
It is an easy way to encrypt Web browsing and also learn more about the TOR network.
Whitney Grace, August 26, 2020
IBM: A New PR Direction without Recipes and TV Game Shows?
August 18, 2020
IBM appears to be shifting its marketing in an interesting way. IBM announced its Power10 chips. Representative of the coverage is Forbes’ Magazine’s “IBM POWER10 Mega Chip For Hybrid Cloud Is Revealed.” The write up is not written by Forbes’ staff. The article is from an outfit called Tirias Research, a member of a contributor group. I am not sure what a contributor group is. The article seems like marketing speak to me, but you judge for yourself. Here’s a snippet:
To handle the ever more complex cloud workloads, the POWER10 improves capacity (socket throughput) and efficiency by about 3x over the POWER9. The energy efficiency gains were critical because IBM increased CPU core count over the POWER9 but kept the socket power roughly the same. All in all, the POWER10 big step forward for the architecture.
Next, I noticed write ups about IBM’s mainframe business. Navigate to “COBOL Still Handles 70% of Global Business Transactions.” The content strikes me as a recycling of IBM-prepared visuals. Here’s an example of the “analysis” and “news” in the article about the next big future:
Several observations:
- It was not that long ago that IBM was touting IBM Watson as capable of matching pets with potential owners. Now IBM is focusing on semiconductors and “workhorse” mainframes
- There are chips using technology more advanced than IBM’s 7 and 14 nanometer chips. Like Intel, IBM makes no reference to manufacturing techniques which may offer more advantages. That’s understandable. But three nanometer fabs are approaching, and IBM appears to be following, not leading.
- The cheerleading for hybrid clouds is different from cheerleading for “the cloud.” Has IBM decided that its future pivots on getting companies to build data centers and hire IBM to maintain them.
The craziness of the state unemployment agencies with COBOL based systems is fresh in my mind. For me, emphasizing the dependence of organizations upon COBOL is interesting. This statement caught my attention:
COBOL still handle [sic] more than 70% of the business transactions that take place in the world today.
Is this a good thing? Are Amazon, Microsoft, and Google embracing mainframes? My hunch is that companies are unable to shift from legacy systems. Inertia, not innovation, may be creating what some people seeking unemployment benefits from COBOL-centric systems perceive as a dysfunctional approach.
Net net: At least IBM is not talking about recipes created by Watson.
Stephen E Arnold, August 18, 2020
-
- Subscribe to Beyond Search
Feature archive
News archive
-
